SENSITIVITY IMPROVEMENT OF A NUCLEAR MAGNETIC RESONANCE METHOD TO MONITOR A BIOARTIFICIAL PANCREAS By NELLY A. VOLLAND A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Nelly A. Volland
3 To Dr. Ioannis Constantinidis (Yanni), suddenly passed away from a massive heart attack the night of April 16th 2007 at the age of 46. He started his project on diabetes over 15 years ago and got me involved in it in 2003. He has left us all orphans after his disappearance. I hope that from wherever he is, he is proud of our progress. And to all the others w ho also left too early, especially Sophie (19791996), and Josephin (1980 2007), we biomedical engineers are working for this never to happen again.
4 ACKNOWLEDGEMENT S The author would like to thank all the people who have made this project possible. The first thanks go to the late Dr. Ioannis Constantinidis for the opportunity, finances, and resources he offered to develop this project, to Dr. Thomas Mareci for taking over the task of becoming her advisor after Dr. Constantinidis death and offering his l aboratory resources and scientific knowledge, and to Dr. Nicholas Simpson for continuing Dr. Constantinidis work and giving her an opportunity to continue the project. The author would also like to greatly thank the members of her committee, Dr. Brennan, Dr. Forder, and Dr. Blackband, and the Department of Biomedical Engineering faculty and staff who have always been present and helpful, especially after Dr. Constantinidis disappearance. The author would also like to greatly thank Barbara Beck and Kelly Je nkins for the time they spent answering the numerous questions regarding electronics and hardware used in the project, discussing the meaning and accuracy of the measurements, and helping with the logistics and good functioning of the NMR equipment. Witho ut them the project would not have gone as far it did. She is also thankful to Dr. Stabler, Dr. Carman, and Dr. Gross for sharing their knowledge on the subjects of tissue -engineered pancreatic construct, material coating, fluorine emulsion, and NMR imagi ng, which are valuable for the presented work. She is thankful as well to the personnel of the Tissue Laboratory, Oca and Chiab, for teaching her everything she needed to learn to successfully culture her own cells, the personnel in the NMR Laboratory, Bi ll, for writing piece of computer code within minutes to analyze imaging data efficiently, and all the undergraduate students who worked with her on the project and gave a tremendous help: Chris, Jess ica, Matt, and Karly. Special thanks also goes to the Advanced Magnetic Resonance Imaging and Spectroscopy Facility personnel for their knowledgeable help every time a problem came and allowing her to
5 keep going even late at night at the facility, as well as to the Animal Care Services employees at the McKnight Brain Institute for their kindness, understanding, and efforts to help with the animal handling, surgeries, and special requests. Special thanks also go to the other graduate students who shared lab space with her, especially Hector, Mansi, Lan, Garrett, and Chelsea for their incredible help and support over the years, their thoughtful discussions, understanding, and funny jokes. Finally, the author would like to thank her family and closest friends John, Alojompo, Ccile, Sid M., Sid P., Anne, Didier, K yle J., and Kennita for all their support and guidance during this work. They have been there every minute of the way, in bad and good moments, always encouraging. Without all of those people and many more, she would have never completed this tedious and complex journey. This work was primarily supported by funding from the National Institute of Health through grant R01 DK47858 and the support of the National High Magnetic Field Laboratory. Partial stipend support was also provided by the J. Crayton Prui tt Family Department of Biomedical Engineering at the University of Florida. This financial support to this interdisciplinary research is greatly appreciated.
6 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ................................................................................................................. 4 LIST OF TABLES .............................................................................................................................. 10 LIST OF FIGURES ............................................................................................................................ 11 ABSTRACT ........................................................................................................................................ 15 CHAPTER 1 INTRODUCTION AND RATIONALE .................................................................................... 17 1.1 Introduction ....................................................................................................................... 17 1.2 Rationale ............................................................................................................................ 20 2 BACKGROUND ......................................................................................................................... 24 2.1 Diabetes ............................................................................................................................. 24 2.1.1 General Facts ......................................................................................................... 24 2.1.2 Diabetes Type I ..................................................................................................... 24 2.2 Tissue Engineered Pancreatic Substitutes ....................................................................... 26 2.2.1 Types of Bioartificial Pancreas ............................................................................ 26 18.104.22.168 Diffusion chamber ................................................................................. 26 22.214.171.124 Microencapsulation ............................................................................... 27 126.96.36.199 Macroencapsulation ............................................................................... 27 2.2.2 InsulinProducing Cells ........................................................................................ 28 188.8.131.52 Transformed -cell lines ....................................................................... 29 184.108.40.206 Genetically engineered stem cells ........................................................ 30 2.2.3 Bioma terials .......................................................................................................... 30 220.127.116.11 Alginate .................................................................................................. 30 18.104.22.168 Agarose ................................................................................................... 31 2.3 Monitoring of Tissue -Engineered Constructs ................................................................. 32 2.3.1 Noninvasive Monitoring In vivo .......................................................................... 33 22.214.171.124 Nuclear medicine imaging .................................................................... 33 126.96.36.199 Optical imaging ...................................................................................... 34 2.3.2 Nuclear Magnetic Resonance ............................................................................... 35 188.8.131.52 Nuclear magnetic resonance phenomenon ........................................... 36 184.108.40.206 Nuclei of interest .................................................................................... 41 220.127.116.11 Nuclear magnetic resonance sensitivity ............................................... 44 18.104.22.168 Radiofrequency coils ............................................................................. 44 22.214.171.124 Coil coating ............................................................................................ 47 2.4 Summary ............................................................................................................................ 51
7 3 DEVELOPMENT OF AN INDUCTIVELY COUPLED RF COIL SYSTEM FOR IMAGING AND SPECTROSCOPIC ANALYSIS OF AN IMPLANTABLE BIOARTIFICIAL CONSTRUCT AT 11.1T ....................................................................... 53 3.1 Introduction ....................................................................................................................... 53 3.2 Background ........................................................................................................................ 53 3.3 Materials and Methods ...................................................................................................... 55 3.3.1 RF Coil System Development ............................................................................. 55 3.3.2 Coating Techniques .............................................................................................. 57 126.96.36.199 Polystyrene ............................................................................................. 58 188.8.131.52 Polytetrafluoroethylene ......................................................................... 58 184.108.40.206 Polymethylmethacrylate ........................................................................ 59 220.127.116.11 Polydimethylsi loxane ............................................................................ 60 3.3.3 RF Coil Testing ..................................................................................................... 61 3.3.4 Samples ................................................................................................................. 64 3.3.5 Const ruct Development ........................................................................................ 65 18.104.22.168 Agarose ................................................................................................... 65 22.214.171.124 Alginate foam ......................................................................................... 65 126.96.36.199 Alginate beads ........................................................................................ 65 3.3.6 Animal Preparation and Handling ....................................................................... 66 3.3.7 NMR Measurements ............................................................................................. 67 3.3.8 SNR Measurement and Statistical Analysis ........................................................ 68 3.4 Results ................................................................................................................................ 69 3.4.1 RF Coil System ..................................................................................................... 69 3.4.2 Implantable Coil Coating ..................................................................................... 74 188.8.131.52 Polystyrene ............................................................................................. 74 184.108.40.206 Polytetra fluoroethylene ......................................................................... 74 220.127.116.11 Polymethylmethacrylate and polydimethylsiloxane ............................ 75 3.4.3 Coil -Construct Assembly ..................................................................................... 79 3.4.4 Phantom Studies ................................................................................................... 80 3.4.5 In vivo Studies in Mice ......................................................................................... 85 3.5 Discussion .......................................................................................................................... 86 3.6 Summary ............................................................................................................................ 90 4 DEVELOPMENT OFA RECEIVE ONLY INDUCTIVELY COUPLED RF COIL SYSTEM TO ENHANCE 1H NMR LOCALIZED SPECTROSCOPY TO MONITOR A N IMPLANTABLE BIOARTIFICIAL CONSTRUCT ............................... 92 4.1 Introduction ....................................................................................................................... 92 4.2 Background ........................................................................................................................ 92 4.3 Materials and Methods ...................................................................................................... 95 4.3.1 RF Coil System Development ............................................................................. 95 4.3.2 Coating Techniques .............................................................................................. 98 4.3.3 RF Coil Testing ..................................................................................................... 98 4.3.4 Samples ................................................................................................................. 99 4.3.5 Cell Culture ......................................................................................................... 100 4.3.6 Construct Development ...................................................................................... 101
8 4.3.7 NMR Measurements ........................................................................................... 102 4.3.8 SNR Measurements and Statistical Analysis .................................................... 105 4.4 Results .............................................................................................................................. 106 4.4.1 RF Coil System ................................................................................................... 106 4.4.2 Implantable Coil Coating ................................................................................... 111 4.4.3 Phantom Studies ................................................................................................. 114 4.4.4 NMR Spectroscopy Studies ............................................................................... 117 4.5 Discussion ........................................................................................................................ 121 4.6 Summary .......................................................................................................................... 127 5 TOWARD THE DEVELOPMENT OF A MULTIPLE -FREQUENCY INDUCT IVELY COUPLED RF COIL SYSTEM FOR SIMULTANEOUS 1H, 19F, AND 31P DETECTION FOR IMAGING AND SPECTROSCOPIC ANALYSIS OF AN IMPLANTABLE BIOARTIFICIAL CONSTRUCT AT 11.1T .......................................129 5.1 Introduction ..................................................................................................................... 129 5.2 Background ...................................................................................................................... 129 5.3 Materials and Methods .................................................................................................... 131 5.3.1 RF Coil Syste m Development ........................................................................... 131 5.3.2 Coating Techniques ............................................................................................ 137 5.3.3 RF Coil Testing ................................................................................................... 137 5.3.4 Samples ............................................................................................................... 139 18.104.22.168 Perfluorocarbons .................................................................................. 140 22.214.171.124 Phosphorus samples ............................................................................. 142 5.3.5 NMR Measurements ........................................................................................... 142 5.3.6 SNR Measurements and Statistical Analysis .................................................... 144 5.4 Results .............................................................................................................................. 145 5.4.1 RF Coil System ................................................................................................... 145 5.4.2 Implantable Coil Coating ................................................................................... 154 5.4.3 Sample Studies .................................................................................................... 155 126.96.36.199 Perfluorocarbon detection ................................................................... 156 188.8.131.52 Phosphorus detection ........................................................................... 158 5.5 Discussion ........................................................................................................................ 159 5.6 Summary .......................................................................................................................... 164 6 CHARACTERIZATION OF A BIOARTIFICIAL PANCREAS AS PART OF AN IMPLANTABLE C OIL -CONSTRUCT ASSEMBLY .....................................................166 6.1 Introduction ..................................................................................................................... 166 6.2 Background ...................................................................................................................... 166 6.3 Materials and M ethods .................................................................................................... 168 6.3.1 Cell Culture ......................................................................................................... 168 6.3.2 Implantable Coil -Pancreatic Construct Assembly ............................................ 168 184.108.40.206 Implantable coil alginate fo am a ssembly ................................ ........... 169 220.127.116.11 Implantable coil alginate beads a ssembly .......................................... 169
9 6.3.3 Cell Entrapment .................................................................................................. 170 18.104.22.168 Alginate foam ....................................................................................... 170 22.214.171.124 Alginate beads ...................................................................................... 171 6.3.4 Culture Propagation ............................................................................................ 172 6.3.5 Glucose Consumption ........................................................................................ 172 6.3.6 Histology ............................................................................................................. 173 6.3.7 Streptozotocin Injection and Blood Glucose Monitoring ................................ 173 6.3.8 Animal Preparation and Handling ..................................................................... 173 6.3.9 Results and Statistical Analysis ......................................................................... 174 6.4 Results .............................................................................................................................. 174 6.4.1 Implantable Coil Construct Assembly Development ....................................... 174 126.96.36.199 Implantable Coil Alginate foam a ssembly ........................................ 174 188.8.131.52 Implantable Coil -Alginate bead a ssembly ......................................... 175 6.4.2 In vitro Study ...................................................................................................... 175 6.4.3 In vivo Study ....................................................................................................... 180 6.5 Discussio n ........................................................................................................................ 186 6.6 Summary .......................................................................................................................... 189 7 CONCLUSIONS AND FUTURE WORK ..............................................................................191 7.1 Conc lusions and Contributions ...................................................................................... 191 7.2 Future Work ..................................................................................................................... 194 LIST OF REFERENCES .................................................................................................................199 BIOGRAPHICAL SKETCH ...........................................................................................................214
10 LIST OF TABLES Table page 3 1 Implantable coil (IC), surface coil (SC), and inductively -coupled implantabl e coil system characteristics. ............................................................................................................ 71 3 2 Implantable coil charac teristics when coated with PS. ........................................................ 74 4 1 Implantable coil characteristic changes aft er the addition of a dec oupling circuit onto the coil. .................................................................................................................................. 108 4 2 Receive -only inductively-coupled implantable coil syste m characteristics. .................... 110 4 3 Comparison of coil characteristics between transmit receive and receive -only c oils and coupled -coil systems. .................................................................................................... 114 5 1 Implantable coil, surface coil, and inductively -coupled implantable coil system characteristics. ...................................................................................................................... 153
11 LIST OF FIGURES Figure Page 2 1 Molecular structure of unbranched polymer chain of alginate ........................................... 31 2 2 M olecular structure of agarose. ............................................................................................ 32 2 3 Example of NMR data. .......................................................................................................... 37 2 4 A SE pulse sequence progra m example to acquire an image. ............................................. 38 2 5 A PRESS pulse sequence program example to acquire a localized spectrum. ................. 39 2 6 A STEAM pulse sequence program example to acquire a localized spectrum. ................. 40 2 7 Molecular structure of Choline. ............................................................................................. 42 2 8 Molecule structure of ATP. ................................................................................................... 42 2 9 Molecule structure of the most c ommonly used perfluorocarbons. .................................... 43 2 10 Molecular structure of PMMA. ............................................................................................. 49 2 11 Molecular structure of PS. ..................................................................................................... 49 2 12 Molecular structure of PDMS. .............................................................................................. 50 2 13 Molecular structure of PTFE. ................................................................................................ 50 3 1 Implantable coil bioartificial pancreas ma croconstruct assembly diagram. .................... 55 3 2 Circuit diagram for an inductively -coupled implantable coil system for 1H detection. .... 57 3 3 Voltage standing wave ratio (VSWR) of a surface coil (SC), an implantable coil (IC), an d an inductively overcoupled system in its different modes. ......................................... 62 3 4 Vector network analyzer outputs use d for RF coils Q measurements. ............................... 63 3 5 Photographs of implantable coils developed as part of an inductively -co upled implantable coil system. ........................................................................................................ 69 3 6 Photograph of the surface coil developed as part of an inductively -cou pled implantable coil system. ........................................................................................................ 70 3 7 RF coil frequency traces, simulated B1 magnetic field magnitude, and their respective corresponding spin echo images along the axis of symmetry of the coil s. ........................ 73 3 8 Photograph of the successfully coated implantable coils. ................................................... 75
12 3 9 Effects of PDMS coating thickness on the imp lantable c oil characteristics. ..................... 78 3 10 Photograph of the implantable coil -construct assemb lies considered in this study. .......... 80 3 11 SNR compa rison between different RF coils tested under unloaded conditions using a water sample. ....................................................................................................................... 81 3 12 SNR comparison between different RF coils tested under loade d conditions in a gel phantom. ................................................................................................................................. 83 3 13 SNR comparison between different RF coils tested under loaded conditions using a gel phantom or in vivo conditions at 11.1T. ........................................................................ 84 3 14 Cross -sectional images of two bioartificial constructs implanted in vivo ......................... 86 4 1 Circuit diagram for a receive -only inductively coup led implantable coil system for 1H detection. ........................................................................................................................... 96 4 2 Implantable coil bioartificial pancreas macroconstruct assembly schematic and dimensions for the transmit receive implantable coil. ....................................................... 101 4 3 Photograph of the RF coi l setup to acquire NMR images. ................................................ 103 4 4 Photographs of the different receive -only implantable coil layouts bui lt and tested. ...... 107 4 5 Photographs of the rece ive -only surface coils tested. ........................................................ 109 4 6 Photograph of a receive -only PDMS -coated implantable coil. ......................................... 112 4 7 SNR comparison between transmit receive and receive -only inductively -coupled implantable coil systems under unloaded conditions using a water sample. ................... 1 15 4 8 SNR comparison between transmit receive and receive -only inductively -coupled implantable coil systems under loaded c onditions using a gel phantom. ......................... 116 4 9 Signal profiles obtained f rom NMR images acquired with TRX and R X inductively coupled systems. ................................................................................................................... 118 4 10 Volume selective imaging acquisition in preparation for 1H localized spectroscopy. .... 119 4 11 1H localized spectroscopy data acquired using a transmit receive inductively -co upled implantable coil system. ...................................................................................................... 120 5 1 Circuit diagram for an inductive ly -coupled double -frequency (1H -31P) implantable coil system. ........................................................................................................................... 134 5 2 Circuit diagram for an inductively -coupled double -frequency (1H -19F) implantable coil system. ........................................................................................................................... 136
13 5 3 Photograph of PFCE emulsion. ........................................................................................... 141 5 4 Implantable coil bioartificial pancreas macroconstruct asse mbly schematic and dimensions. ........................................................................................................................... 141 5 5 Voltage Standing Wave Ratio (VSWR) of the double frequency 1H -31P inductivelycoupled coil system at both frequen cies. ............................................................................ 145 5 6 Near magneti c field distribution along the axis of s ymmetry of the double -frequency 1H -31P inductively -coupled coil syste m at 190.5 MHz and 470.75 MHz. ........................ 146 5 7 Voltage Standing Wave Ratio (VSWR) of the do uble frequency 1H -19F inductivelycoupled co il system at both frequencies. ............................................................................ 148 5 8 Near magnetic field distribution along the axis of symmetry of the double frequency 1H -19F inductively -couple d coil system at 442.86 MHz and 470.75 MHz. ...................... 149 5 9 Photograph of the double -frequency 1H -31P implantable coil. .......................................... 149 5 10 Photogr aph of the double -frequency 1H -31P surface coil. ................................................. 150 5 11 19F images of a pancreatic construct surrounded by an implantable coil and containing BaA beads with 2.5 % PFCE. ............................................................................ 157 5 12 19F spectrum series from BaA beads containing 0.4% PFCE within a pancreatic construct surr ounded by an implantable coil. ..................................................................... 158 5 13 31P spectru m of a saline solution containing 500 M of ATP and PO4 2. ........................ 159 6 1 Photograph of an implantable coil alginate foam construct assembly prior to being soaked in PBS. ...................................................................................................................... 169 6 2 Implantable coil -bioartificial pancreatic construct as sembly diagram and dimensions. 170 6 3 tet cells encapsulated in alginate beads prio r to insertion in a construct. ............................................................................................................................... 175 6 4 Temporal changes in the glucose consumption rate (GCR) of TC tet cells enca psulated in AP A and BaA beads. ................................................................................. 177 6 5 Hematoxylin/eosin stained cross -sections from BaA beads prepared by encapsulating TC tet cells in 2% (w/v) LVM alginate gelled with 55 mM BaCl2. ....... 179 6 6 Hematoxylin/eosin stained cross -sections from APA beads prepared by encapsulating TC tet cells in 2% (w/v) LVM alginate gelled with 100 mM CaCl2. ..... 179 6 7 Hematoxylin/eosin stained cross -sections of the tissues found inside and around the construct after its retrieval f rom a mouse peritoneal cavity. ............................................. 181
14 6 8 MR images of pancreatic substitutes implanted in the peritoneal cavity of two differe nt mice. ....................................................................................................................... 181 6 9 MR images of cell -free beads within t he construct cavity over time. ............................... 182 6 10 Photog raphs of a piece of PEEK mesh. .............................................................................. 182 6 11 Temporal changes in diabetic mice blood glucose levels after implantation of TC tet ce lls encapsulated in BaA beads. .......................................................................... 184 6 12 Hematoxylin/eosin stained cross -section from implantable coil -construct assembly cavity tissue after 21 days of implantation in a mouse peritoneal cavity. ........................ 185
15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SENSITIVITY IMPROVEMENT OF A NUCLEAR MAGNETIC RESONANCE METHOD TO MONITOR A BIOARTIFICIAL PANCREAS By Nelly A. Volland August 2009 Chair: Thomas H. Mareci Major: Biomedical Engineering Non -invasive monitoring of implanted devices is becoming key in develop ing tissue engineered structures intended to provide alternat ive or complementary treatment to people with organ or tissue impairment and/or loss. B ioartificial pancreas es for the treatment of Type I diabetes are an example of such structures under development. Nuclear magnetic resonance (NMR) imaging and spectros cop y have already shown their potential for monitoring of these pancreatic substitutes and detect ing their early marker of failure However the sensitivity of th ese NMR methods was limited The studies presented in this dissertation investigate sensitiv ity improvement by developing inductively -coupled implantable RF coil systems at high magnetic field strength not only for 1H detection, but also for the detection of other informative, but less sensitiv e nuclei, such as 19F and 31P. An inductively-couple d implant able coil system was first developed for 1H detection to demonstrate the use on inductively coupled implant able coil system at 11.1 T Secondly, a development of a receive -only inductively-coupled implanted coil system was investigated to further improve 1H detection and increase localized spectroscopy performance. The feasibility of double frequency inductively -coupled implantable coil systems for simultaneous detection of 1H -31P and 1H -19F were explored next, toward the development of a multipl e frequency system. T he requirements of th e s e coil system s for a complete monitoring of
16 a bioartificial construct were discussed In parallel to the coil system developments, the inclusion of an RF coil within the implanted pancreatic construct was also studied to address the restrictions it imposed on the construct design. The results for 1H detection establish that large gains in signal to noise can be obtained with the use of inductively -coupled implanted coil system s w hen compared to the use of standard surface coil s This improvement provides a means to better analyze the structure s of implanted bioartificial construct s and their changes over time on NMR images. It can also lower the limit of detection when spectroscopy is performed to detect choli ne NMR signal and allow for better quantitative analysis of bioartificial organ functions. A receive only inductively -coupled implanted coil system was also successfully developed to further enhance localized spectroscopy since a more homogene ous NMR exci tation magnetic field could be achieved with its use compared to the use of a transmit -receive system 31P and 19F spectroscopy was also performed using single -frequency inductively -coupled implanted coil system for 31P detection to detect adenosinetriphosphate (ATP) and for 19F detection to detect perfluorocarbons (PFC). These results set the standards for the development of double -frequency inductively -coupled implantable coil systems. The feasibility of double frequency inductivelycoupled implantable coil system was assessed as well Furthermore, data shows that a coil -construct assembly allowed insulin producing cells to function and stay viable for extended periods of time in vitro The return to normoglycemia in diabetic mice after coil construct assembly implantation was also demonstrated while enhanced non invasive monitoring of the implanted constructs was made possible using NMR methods.
17 CHAPTER 1 INTRODUCTION AND RAT IONALE 1 1 Introduction Non -invasive monitoring of implanted devices is becoming key in t he development of tissue -engineered structures to provide alternative or complementary treatment to people with organ or tissue impairment and/or loss. Tissue -engineered pancreatic substitutes for the treatment of Type I diabetes is one example of tissue -engineered structures that would largely benefit from such a direct in situ monitoring method. Bioartificial pancreatic substitutes or constructs implanted to compensate for the loss of the natural insulin-producing cells ( -cells) occurring in type I dia betes have shown promise toward restoring normoglycemia in diabetic animals with near normal regulation over an extended period of time (1). These constructs are made of semi permeable membranes encapsulating insulin-producing cel ls. The encapsulated cells respond directly to changes in their surrounding glucose level by producing insulin. This highly regulated system offer s better blood glucose regulation than the most common and most accessible treatment consisting of regular b olus insulin injections throughout the day. Currently, therapeutic efficacy is assessed by the successful restoration of the hosts glucose regulation. Although measuring the hosts blood glucose can indicate the construct status, thi s is an indirect tech nique that only evaluate s the response of the host to the tissue -engineered construct functions and performances To fully understand the function of such constructs, optimize their design and development, and predict their behavior over time, their full characterization, quality control after development, and performances and life span in a given environment are needed Histology is a common method for such characterization, but is destructive, and invasive. However, i n order to assess changes in constr uct function that may be predictive of co nstruct failure well in advance of hyperglycemia, a direct and non -invasive method to monitor the
18 constructs in vivo post implantation over time is desired. A method well suited to perform the direct and noninvasi ve in vivo monitoring of a pancreatic construct is nuclear magnetic resonance (NMR) (2 4) NMR imaging and spectroscopy can repetitively collect structural and functional information from these constructs simultaneously without altering them or their composition Recent studies utilizing NMR to non -invasively monitor tissue -engineered pancreatic constructs were encouraging (5,6) However the sensitivity of the monitoring system used in these studies was limited by the strength of the magnetic field and the radiofrequency (RF) antennas (or coils) used to detect the 1H NMR signal It further prevented the detection of informative but less sensitive nuclei (compared to 1H ), such as 19F and 31P. The endeavor of this thesis was to improve the sensitivity of the NMR method mentioned above to non -invasively monitor a pancreatic construct in vitro and in vivo and identify early markers of construct failure Efforts were specifically focused on using higher magnetic fi eld strengths and enhanced RF coil s to improve the sensitivity of the NMR method used in the previous studies by at least a factor of four. Higher magnetic fields allow a higher net magnetization of the sample which increases the sensitivity of the method (7). Modifying the RF coil design by matching its size, shape and proximity to the structure of interest can enhance the sensitivity of the MR detection as well as the selectivity of the MR signal (8). The approach taken in this research was to develop inductively-coupled RF coil systems to increase the sensitivity of the NMR methods establish ed previously (5,6) to non invasively monitor a bioartificial pancreas at 11.1 T in vitro and in vivo Th e coil system includ es a coil surround ing the tissue -engineered construct (implantable coil) to be monitored. Implantable coil systems have been developed and successfully used over the past 25 years to increase the sensitivity o f NMR detection (9,10) Inductively -coupled implantable coil system s have
19 appeared more recently to eliminate the external cable connection to the implantable coils which encouraged infection and discomfort at the site of cable exit (11 13) Therefore inductivelycoupled implantable coil system s w ere investigated to improve the sensitivity of NMR imaging and spectrosc opy toward monitor ing non invasively a tissue engineered construct. In this study, the tissue -engineered model of choice was a bioartificial pancreas. Coil system prototypes, including an implantable coil inductively -coupled to a surface coil, w ere designed for 1H detection and characterized on the bench as well as in vitro and in vivo The improvement of the detection sensitivity of the NMR imaging and spectroscopy methods was a primary concern. The effects of different coil types and differen t implantable coil coatings were investigated and quantified to determine the system with the best sensitivity improvement in vivo compared to a surface coil. Furthermore, receive -only inductively coupled implantable coil systems were later developed and investigated to evaluate their localization improvement when used for 1H magnetic resonance spectroscopy (MRS). The feasibility of m ultiple -frequency inductively coupled implantable coil systems were also explored since it would allow the simultaneous det ection of different nuclei than 1H, such as 19F and 31P which would help in the full characterization of the bioartificial pancreas in vitro and in viv o While a 1H -31P and a 1H -19F inductively-coupled implantable coil systems were studied, their single -frequency counterparts for 19F and 31P detection were also developed to assess the ir efficacy and set the standards for the multiple -frequnecy coupled system performances Different types of bioartificial pancreatic construct were also studied in parallel of the coil system development to define the most appropriate construct to perform its diabetes regulation duties while allowing its monitoring using NMR methods.
20 1 2 Rationale This dissertation consists of four distinct topics presented in Chapters 3 6. Som e of these chapters represent a modified version of an article published or submitted for publication. The individual chapters all include a brief introduction, a focused background, a material s and method s section, a presentation of results obtained a d iscussion and a brie f summary Chapter 2 presents the general background information of this research. The use of inductively-coupled implantable coil system s to improve the sensitivity of NMR imaging and spectroscopy methods to noninvasively monitor a bioartificial pancreas required a methodical characterization of the RF coils. Therefore, the approach here was four fold: (1) the d evelop ment and characterization of an inductively-coupled single -frequency implantable coil system for 1H detection and the test ing of its sensitivity in vivo ; (2) the d evelop ment and characterization of a receive -only inductively-coupled single -frequency implantable coil system to further improve the localization of the NMR signal within the construct when performing 1H locali zed spectroscopy ; (3) the evaluation of the feasibility of multiple -frequency RF coil system s for the simultaneous detection of 1H, 31P, and 19F ; and (4) the development and characterization of the construct in which the implantable coil of an inductively-coupled coil system was inserted While several different implantable coil designs were explored in this research, all systems used a similar surface coil as their power source. Initial studies were made to design and select the best inductively -coupled i mplantable coil system for 1H detection compared to a surface coil. Different implantable coils and configurations were considered. Different coatings for the implantable coil were also evaluated, including polymethylmethacrylate (PMMA) (14) and polydimethylsiloxane (PDMS) (15 18) Numerous studies have been published on inductively-coupled implantable coil systems (14,1921) ; however, Chapter 3 discusses the development of such a system for monitoring a
21 bioartificial construct in vitro and in vivo demonstrates its use at high field (11.1 T), and investigate s its requirement s for it s use in such an environment While an inductively -coupled implantable coil system provide s the sensitivity improvement needed to better analyze the structure and function of implanted bioartificial organs the RF magnetic field homogeneity it creates coul d be further enhanced to increase the localization of 1H MR spectroscopy (MRS). Chapter 4 details the development and the characterization of a receive only inductively-coupled implantable coil system and demonstrate s its use at high field (11.1 T) D eve loping the inductively-coupled receive -only implantable coil system was easily made by the addition a decoupling circuit to each coil of the system Howver, the receive -only implantable coil coating, the inductive coupling of a receive-only load ed system and the magnet testing of this receive -only system required further consideration than the transmit receive system presented in Chapter 3. If 1H detection through imaging and localized spectroscopy is important for monitoring a tissue -engineered construct, such as a bioartificial pancreas, phosphorus (31P) and fluorine (19F) detection s are essential to bring additional information about the construct well being and functioning and achieve the construct complete monitoring. Chapter 5 describes the requireme nts and use of single -frequency inductively -coupled implantable coil system for 31P and 19F detection a s well as the feasibility of two double -frequency inductively -coupled implantable coil systems, (1H -31P & 1H -19F) The single frequency systems were cha racterized to assess the detection sensitivity and to determine afterwards, the losses multiple -frequency systems suffer for the 1H 31P and 19F simultaneous detection. The behavior of the i mplantable coils at different frequencies when coated and loade d was also studied to observe the change s
22 occurring under different conditions This is a significant consideration for the development of double -frequency inductively -coupled implantable coil systems. Once the use of the different inductively coupled implantable coil systems demonstrated the sensitivity improvement obtained using NMR methods to noninvasively monitor a bioartificial pancreas the inclusion of the implantable RF coil within the pancreatic construct was studied to characterize the restrictions it placed on the construct design and function. Chapter 6 discusses the development of two flat disk macroconstructs including an implantable coil It describes the steps needed to allow their design to incorporate a coil and its coating and allow th e non -invasive monitoring of implantable constructs using NMR methods. T he coil construct assembly effects on the entrapped TC -tet cells viability and function and its ability to reverse a streptozotocin -induced diabetic state in mice were studied. Pre treating the assemblies was also investigated to prevent construct fibrotic overgrowth in vivo in a mouse peritoneal cavity Chapter 7 summarize s t he implications of the findings presented here on the use of inductively-coupled implantable coil system for 1H, 19F, and 31P detection to perform N MR imaging and spectroscopy and monitor tissue engineered constructs, specifically a bioartificial pancreas The restrictions the implantable coils have on the construct development and the evolution of the inductiv ely -coupled implantable coil systems in future research are also discussed in this last chapter The aim of this project is to develop highly-sensitive noninvasive NMR techniques to monitor a bioartificial pancreatic construct in the treatment of diabetes type I This monitoring potential is critical, as it may allow for the identification of early markers of construct failure long before functional failure of the construct (i.e., hyperglycemia) easily detected by blood glucose measurements. The reason f or this optimism is through analogy to the normal pancreas.
23 T h ere are more insulin-secreting cells ( -cells) in the body than needed to maintain normoglycemia I n the intial stage of type 1 diabetes the number of -cells is dwindling The physiological effects of this cell loss (hyperglycemia) are not experienced until 5070% of the se cells are dead. Likewise, we expect that if we also have an excess of insulin -secreting cells in a bioartificial pancreatic construct, failure of the construct may occur o nly after a significant fraction of the original total have died. Through careful monitoring, once the lower limit of cell number required to maintain normal function has been determined, the construct can be replaced well before that level is reached, th us preventing hyperglycemia, and effectively controlling diabetes with the implantable construct.
24 CHAPTER 2 BACKGROUND 2 1 Diabetes 2 1 1 General F acts Diabetes mellitus, insulin and noninsulin dependent also known as type I and type II diabetes respectively, is a yet incu rable endocrino logic disease that affect ed 2 4 million children and adults in the United States, about 7 .8 % of the population in 2008. Between 5 to 10% of diabetic people are diagnosed with type I Diabetes (22) Diabetes is one of the fastest growing diseases where blood glucose levels become abnormally high and which c an lead to major secondary health consequenc es. According to the American Diabetes Association, it currently costs an estimated $ 174 billion in health care expenditures. 2 1 2 Diabetes Type I D iabetes Type I results from an autoimmune disease that destroys the -cells residing within the islet s of Lange rhans in the pancreas thu s diminishing insulin production (23) As the insulin secretion decreases, blood glucose levels cannot be sustained within the physiological ra nge (80 120 mg/dl). The most common and most accessible treatment to compensate for the need of functional insulin is regular injection of exogenous insulin subcutaneously throughout the day via a syringe. T his bolus type of delivery does not offe r optim al blood glucose control making h igh or low blood glucose values difficult to avoid. The p atients average blood glucose level is usually higher than the physiologic average with this type of treatment This standard management of diabetes offers patient s a near normal life, but requires a close and constant monitoring of the disease. One alternative to this treatment is the use of an insulin pump via a catheter placed under the skin in the peritoneal cavity. This setup allows for continuous delivery of a low base level of insulin to the body throughout the day in addition to bolus doses similar to
25 the ones obtained with syringe use. If the blood glucose level is on average lower and better regulated when using the insulin pump than with the use of syri nge injection alone, it still does not optimally control the blood glucose level. The large amount of glucose in the blood vessels can attach to proteins and alter their structure and function. As a consequence, blood vessels tend to become thicker and s tiffer impairing blood flow and leading to more important complications, such as heart disease, nephropathy, retinopathy, and limb amputation. Even though the death rates of people with type I diabetes is much lower now than 50 years ago because their blo od glucose levels are better regulated and more physiologic, 7% of them die within 25 years of diagnosis. Furthermore, their life expectancy is still reduced (24) due to the major health consequences associated with the disease and which cannot b e prevented despite vigilance. Pancreas transplantations have been increas ingly performed over the past forty years in patients exhibiting advanced secondary complications and inadequate glucose control despite insulin therapy (25,26) Pancreatic islet transplantations have also been studied (26,27) Even though improved isolation, transplantation implantation methods and new immunosuppression regimen have been developed (28) this procedure is still greatly limited by the insufficient supply of donor tissue. Often two cadavers are needed to isolate the required number of islets for the transplantation in one patient. Therefore, this treatment is currently still reserved for people diagnosed with Diabetes type I with inadequate blood glucose level control despite insulin therapy. Furthermore, a life long immunosuppressant treatment is required after the transplantation to prevent the rejection of the dono r tissue These limitations have been the motivating factor for the development of an effective treatment which can provide physiologic
26 blood glucose regulation without the use of an immunosuppressive treatment. Tissue -engineered pancreatic constructs pr ogressively appear to fulfill these requirements. 2 2 Tissue -Engineered Pancreatic Substitute s Tissue -engineered pancreatic constructs, also called bioartificial pancreases (BAP), emerged over twenty -five years ago. They were first developed to improve islet survival after transplantation without the need for immunosuppressive therapy (1). However, with the limited availability of pancreatic islets, other sources of insulin producing cells had to be found. T he different types of BAP that have mainly been investigated thus far include diffusion chambers, microencapsulation, and macroencapsulation. These devices can be inserted into the blood supply or implanted anywhere in the body that has diffusion potential. The y are comprised of individual cells or isletlike clusters of cells within a biocompatible semipermeable membrane which provides mechanical protection and partial immunoisolation. Optimally, the membrane should be permeable to low molecular weight nutrients and metabolites, but impermeable to higher molecular weight molecules, such as immunoglobulin and albumin. 2 2 1 Types of Bioartificial Pancreas 2 2 1 1 Diffusion c hamber The diffusion chamber was the first BAP shown to restore normal glycemia in animals over extended periods of time (29) It can be made of one or several tubular semip ermeable hollow fiber membranes The se devi c es a re commonly referred to as intravascular devices because they are inserted as an artery to -vein (AV) shunt in the hosts body (30,31) The cell s or islets are embedded on the outside surface of the fibers and t he blood circulat es inside the fiber s The closeness of the islets to the blood flow allows them direct acces s to nutrients and oxygen to maintain islet viability and functionality. High diffusion of glucose through the chamber provides immediate recognition of glucose level changes and triggers h igh diffusion and
27 immediate release of insulin H owever, the surg ical implantation of the device is relatively risky and bl ood coagulation in the fibers has been shown to be a major issue with this intra vascular device (29,32) Because of significant b lood flow changes in the proximity of the devi c e it was quickly abandoned. Due to these issues, t he development of extravascular devices, such as micro and macroencapsulated devices was initiated T hese latter device s present major advantages such as smaller size, flexibility in selecting implantation site, and minimal surgical risk However, these advantages come at the cost of significant diffusion limitations. 2 2 1 2 Microencapsulation In microencapsulated devices, the c ells or islets are suspended in a polymeric material, gelled, and enclosed within a biocompatible semipermeable membrane to produce microspheres or beads with diameter s from a fraction of a millimeter to a couple of millimeters (33) These capsules are small enough to be implanted almost anywhere in the body with a simple needle. The liver and the spleen were initially preferred implantation locations due to their higher diffusion capabilities and their access via th e portal vein (34) Despite its limited diffusion capabilities (e.g. lower vascularization), t he peritoneal cavity has subsequently become the main implantation site (33, 35,36) because of simple access ibility and holding capacity. Although microencapsulation has been widely used in the development of BAP (35,37) one drawback is the wide dispersion of the implanted capsules throughout the body. This dispersion makes in vivo monitoring and retrieval of this type of BAP nearly impossible. 2 2 1 3 Macroencapsulation Macroencapsulati on involves generating a construct capable of containing a large number of cells. The construct architecture encompasses a matrix, where the cells are suspended between two semipermeable membranes. These devices can take the shape of a tube or a planar
28 s heet and are generally implanted in either the peritoneal cavity or a subcutaneous location due to their size and geometry. The tubular or fiber macrocapsules are very similar to diffusion chambers used as vascular devices. The main difference resides in the fact that the cells, islets, or microcapsules are embedded inside the fiber instead of on the outside surface (38) Flat sheet membranes are often large circular (39,40) or rectangular (41) sealed chambers in which the cells are immobilized. These devices allow for (i) cell viability studies, (ii) membrane biocompatibility, diffusional, and immunogenic assessment studies, and (iii) insulin and glucose transport modeling. Both micro and macroencapsulation techniques will be considered here to generate BAP with high diffusion capability, and keep the entrapped cells all together in one location. A combination of these techniques w ill also be investigated to make t he entire implant easy to retriev e and to monitor closely in vivo 2 2 2 InsulinP roducing C ells A major component of all BAPs is the insulin-producing c ells. Insulin is normally produced by the -cells of the Langerhans islets of the pancreas. Ideally, these cells would be used in transplant constructs. However, islets only constitute 12% of the total mass of the pancreas, thus their availability is limited. Moreover, one can only obtain islets from dead individuals, so the demand far outweighs the available supply. Furthermore, their isolation remains a tedious process in which more than one cadaveric pancreas is usually required to obtain sufficie nt functional islets to treat a patient. Unfortunately, their proliferation in culture has been proven to be difficult and not yet fully successful (42) These limitations strongly encouraged the development of modified insulin-producing cell lines. To this end, cell lines from either or non origin and from either human or non -human source have been generated
29 as possible human islet substitutes since they can easily and indefinitely be amplified in vitro Many cell lines are currently available, such as porcine islets, transformed -cell lines (e.g. TC3 and TC tet line ), and genetically engineered stem cells. 2 2 2 1 Transformed -cell l ines TC3 cells are insulin -producing tumor cells generated from in sulinomas derived from transgenic mice carrying the hybrid insulin promoter -simian virus 40 antigen gene (43,44) The major difference between these cells and normal pancreatic islet -cells is that they do not exhibit a normal insulin response to glucose stimulation. They ca n be triggered by lower levels of glucose to produce insulin, but immediately reach a maxim al insulin release rate. Additionally, they do not include all the regulatory components of the normal -cells. However, their lower nutrient and oxygen requiremen t to maintain normal secretory function is an advantage when encapsulated in BAPs. T he islet insulin production stops when the oxygen partial pressure goes below 5 mmHg (45,46) whereas the TC3 cells keep an unaltered insulin production until the oxygen partial pressure becomes 7 mmHg (47) Their proliferative capacity also provides a virtually unlimited supply of these insulin-producing cells. However, this becomes a liability in vivo as their continuous proliferation will eventually pro duce enough insulin -producing cells to put animals and/or humans in danger of hypoglycemia. To alleviate this problem, cell lines with an integrat ed bacterial tetracycline operon regulatory system (tet) have been developed to regulate cell proliferation. TC tet cell line incorporates a tet -off system that stops TC cell proliferation when tetracycline (or a derivative) is present in the cells environment (48) This latter cell line will be used in this project even though non cell lines can also be genetically engineered to secrete insulin.
30 2 2 2 2 G enetically -e ngineered stem c ells Genetic engineering concept has been demonstrated using hepatocytes (49) fibroblasts, and enteroendocrine cells (50,51) The main advantage of noncells is that they can be derived from an autologous source, thus reducing the chances of immune response. Furthermore, s tem cells, either embryonic (52) or adult/progenitor (53) have also shown the ability to form isletlike clusters of insulin -producing cells (54) However, noncells and stem cells are still in their developmental phase and hence will not be considered in the presented work 2 2 3 Biomaterials The other major component of any pancreatic construct is the material used to contain the cel ls and to generate the semipermeable membrane. Several materials can be selected based on their biocompatibility, structural integrity, and diffusion properties These factors are important to consider since they impact the viability and function of the entrapped cells. Ease of manufacturing and encapsulation conditions are other criteria to take into consideration when choosing the membrane material. The most commonly used materials in studies of both micro and macroencapsulation BAP designs are algin ate (33,37,55) and agarose (39,40,56,57) 2 2 3 1 Alginate Alginates ar e unbranched polymers found in a wide variety of brown seaweeds and some bacteria. They are comprised of L D -mannuronic acid (M) segments in varying proportions, order, and molecular weight (Figure 2 1). These variables dictate the physical properties of the alginate hydrogel generated in the presence of multivalent cations, such a s calcium barium and s trontium Alginates with a high G content tend to form more rigid gels at a very specific cation concentration, whereas higher M segment ratio alginates tend to form a softer gel over a wider range of conditions (58) The divalent cation bonding to the G segments rather than the M segments is the reason for the difference in gel structure. The
31 different cation ionic bonding influences the physical properties of the gel (59). This flexibility in producing alginate gels or foams of various stre ngth s offers the ability to create the cell matrices and semipermeable membranes needed in tissue engineered constructs (60,61) Therefore, two types of alginate ( gel a nd foam ) will be considered in the presented work for both micro and macroencapsulation Calcium (62) and barium (59) will be also te s t ed as cationic bonding agents. Figure 2 1. Molecular structure of unbranched polymer chain of alginate This molecule contain s both the -L -guluronic acid (G) segment (right) and D -mannuronic acid (M) segm ent (left) randomly ordered. 2 2 3 2 Agarose Similarly, agarose is a linear polysaccharide isolated from agar found in marine red algae. It is made of basic repeating agarobiose units, comprised of alternating segments of D galactose and 3,6 anhydro L -galactos e (Fig ure 2 2). It can vary widely in molecular weight. Agarose gels are formed by suspending dry agarose in aqueous buffer, then boiling the mixture until a clear solution forms ( above 90 C) (63) The gel forms as the solution cools. Unlike alginate, this gel is weakly ionic with a large distribution of pore sizes depending on the agarose concentration (64)
32 Figure 2 2. Molecular structure of agarose This molecule is composed of agarobiose units of D -galactose and 3,6 anhydro L -galactose. A garose will be explored here to generate macrocapsules. Other materials that have also been studied as cell matrices and/or semipermeable membranes are polyvinyl alcohol (PVA) (41,65) and poly(N,N -dimethyl acrylamide) or PDMAAm crosslinked by hydrophobic polyisobut ylene (PIB) moieties (PDMMAm l -PIB) (66) and sili cate (36,67) Even though t hese materia ls have shown promising results, they are still in their early developmental stage and will not be considered in the presented work 2 3 Monitoring of Tissue -Engineered Constructs In vitro and in vivo monitoring is of primary importance for the development of tissue engineered construct s as it can be a tool to study the construct functionality temporally and optimize its design and function. Measuring a subjects blood glucose and insulin concentrations is currently the main technique to evaluate the behavior of the cells imbedded within the semipermeable membrane of a BAP in vivo These indirect techniques evaluate the response of the host to the BAP functions and performances in vivo Unfortunately, they cannot provide information about the mechanical integrity of the construct nor its potential demise. Histology can assess the constructs integrity and its mechanisms of failure (if any). However, it is a destructive, non -dynamic, and non real time technique. Non -invasive dynamic and real time methods for monitoring construct cell viability and metabolism can provide the time -dependant
33 information of experimental, developmental, environmental, and therapeutic influences on the construct in the same subject. It also allows the realization of practical and c linically relevant constructs. 2 3 1 Noninvasive Monitoring In v ivo Several methods are available to directly monitor tissue -engineered constructs noninvasively in vivo Each of them differs from each other in its spatial and temporal resolution, depth penetr ation, image generation mode, molecular probe need and availability, specificity, and sensitivity (68) Positron emission tomography, single photon emission computed tomography (69,70) optical microscopy (71) and nuclear magnetic resonance imaging and spectroscopy (2 4) have been the most widely investigated monitoring methods as diabetes treatment modalities. 2 3 1 1 Nuclear medicine imaging Positron e mission tomography (PET) and single photon emission computed tomography (SPECT ) have already been shown to be safe and feasible techniques to image transplanted pancreatic islets non invasively (69,70) Though SPECT is less commonly used, both techniques can be used to image tissue -engineered constructs noninvasively and track cellular function and viability. PET generates images by recording high -energy -rays emitted from radionuclide probes introduced in the subject. N atural biological molecules can be labeled with an isotope, such as 15O, 13N, 11C, and 18F (substitute for 1H), capable of producing a positron from its nucleus, which will eventually annihilate with a nearby electron to create two rays needed for PET (72) DeLonlay et al. used 18Fluoro-L dihydroxyphenylalanine (DOPA) to evaluate the difference between diffuse and foca l hyperinsulinism (70) as pancreatic -cells can take this radionuclide up and convert it into dopamine. Cells can also be engineered to express a reporter
34 gene that will allow the uptake of a specific radionuclide. Lu et al. created cells expressing the herpes simplex virus 1 thymidine kinase (HSK TK) gene to al low the uptake of labeled 9 (4 fluoro 3 -hydroxymethylbytyl)gua n ine (FHBG) (69) to monitor transplanted islets after implantation SPECT employs the detection of rays from -emitting isotopes, such as 99 mTc, 111In, 123I, and 131I to generate images. Unfortunately, t he in jection of these radionuclide probes or the modification of the cells of interest to express a PET or SPECT re porter gene are two major drawbacks of these techniques. Furthermore, the half -life of the radionuclide probes is relatively short Even though these techniques have no depth penetration limitation and their sensitivity and specificity are high (only 1010 to 1012 mol/L radionuclide is needed to generate images), t he best resolution they can achiev e is only 1 2 mm Simultaneous multiprobe detection is also impossible in PET, although SPECT could have some capability of using isotopes of different energy 2 3 1 2 Optical imaging Optical imaging techniques have become of more and more interest recently with the advancement in silicon -based charged coupled devices (CCD) detectors. Bioluminescence imaging (BLI) uses the emitted light from a living structure result i ng from a chemical reaction during which chemical energy is converted to light energy to generate an image Often this requires the modification of the cells of interest to express a bioluminescence reporter gene. Fowler et al. created an adenovirus encod ing firefly luciferase to express luciferase in pancreatic islets and monitor them after transplantation in mice and measure their mass over time (73) T he detectors also need to be cooled in order to reduce thermal noise and increase the imaging sensitivity. Fluorescence imaging utilizes the emissio n of light triggered by the molecular absorption of a photon from an external light source with a different wavelength to generate an
35 image. The introduction of fluorescent probes such as fluorophores and green fluorescent proteins (GFP), is often necess ary. Quantum dots, nanoscale semiconductor crystallites with the ability to emit monochromatic light at a specific wavelength and color, are another possibility (74,75) These techniques allow real time, non -invasive and non-destructive imaging of living tissues at a very high spatial resolution (micron and submicron level ). Confocal microscopy (CM) is one popular fluorescence -mediated techniq ue that ha s enabled high contrast three dimensional imaging of tissue structures in intact, optically non transparent samples. However, th e introduction of fluorescent probes with limited lifetime in the sample (71) to enhance the imaging contrast is one major limitation of this technique Relatively intense light power may also be required to obtain adequate image quality. These two requirements are often detrimental to cell viability. Furthermore, the limited light penetration depth (only a few hundred microns) is one more issue with this modality as constructs are implanted well beneath the skin surface (a couple of millimeters deep) The use of m ultiphoton microscop y instead of confocal microscopy can improve imaging depth to some degree However, a n ear infrared ( NIR ) optical imaging technique like optical coherence tomography (OCT) can be used to overcome several limitations in optical imaging (76) It measures the intensity of back reflected light and relies on the variations of indices of refraction of optica l scattering for image contrast. Therefore, no fluorescent probe is needed. N ear -i nfrared light also has a greater penetration depth compared to light used to excite fluorescent probes Nonetheless, this optical technique offers most of its advantages to in vitro studies because the imaging depth can never exceed two or three millimete rs. 2 3 2 Nuclear Magnetic Resonance N uclear Magnetic Resonance (N MR ) has also been used to monitor non invasively tissue engineered constructs in real time (77) This technique can be applied to a variety of living
36 systems ranging from a si ngle cell (78) to an entire human being (79) and can yield both chemical information (spectra) or structural information (images). It can be implemented to provide excellent contrast in soft tissue with high resolutio n (< 100 m), and can also detect a variety of important metabolites deep with in a sample. NMR techniques represent the best compromise to monitor a BAP longitudinally and non -invasively in vitro and in vivo (2 4) 2 3 2 1 N u clear magnetic resonance phenomenon The NMR phenomenon takes advantage of nuclei that have a nuclear spin. In presence of a magnetic field, referred as to the main or static magnetic (B0), these nuclei align with or against th is magnetic field creating a known state of equilibrium in the sample The difference between the two nuclei (spins) populations results in a net magnetization of the sample (M0) along the field (B0). These spins have an associated resonan t frequency (f0, in the radiofrequency (RF) range) at which they can interact. Because of this property, specific RF waves or pulses can be applied to a sample to generate an excitation magnetic field (B1) perpendicular to B0 and put these spins in an excited state. These RF pulse s are generated by RF antennas, also called transmit RF coils. An excitation or 90 pulse cancels the sample net magnetization (M = 0), whereas a refocusing or 180 pulse inverse it (M= -M0) After the RF pulse, the relaxation of the spins back to equilibrium creates RF magnetic fields that can be detected by a receiver RF coil which generates an RF signal called Free Induction Decay (FID) This return to equilibrium occurs due to various mechanisms. It can be measured as the recovery of the magnetization along B0 (de fined as the Z -direction), and the loss of magnetization in the orthogonal (XY) plane after B1 ended. These components exhibit exponential decay dependence defined as longitudinal or spin lattice (T1) and transversal or spin-spin (T2) relaxation times, re spectively. Magnetic gradients can be superimposed upon the main magnetic field to localize the FID to a specific
37 volume of interest (VOI) in the three -dimensional object being studied. This is the basis of the formation of both NMR images and spatially localized NMR spectra (Fig ure 2 3). Figure 2 3. Example of NMR data. A ) I mage of a bioartificial pancreatic construct in vivo ; and B) 1H NMR spectrum of a 1 mM choline solution in D2O displaying the different resonances of choline. For magnetic resonance imaging (MRI), a slice encode gradient determines the orientation of the slices acquired. Phase and frequency encode gradients, the latter often being called read out gradient, are applied perpendicular to the slice gradient to allow the signal acqui sition pixel by pixel and the generation of 2D images (80) after applying Fourier transform s (FT) The combination of t he gradients and RF pulses creates a pulse sequence. A commonly used imaging sequence is the s pin -echo (SE) sequences where a n excitation and a refocusing pulse are combined with slice, phase encode, and read out gradients (Fig ure 2 4 ). By varying times between the different pulses (repetition time (TR) and time of echo (TE)), the images (Figure 2 3A) can be T2-weighted, T1-weighted, or representing the spin density of the sample.
38 Figure 2 4. A SE p ulse sequence program example to acquire an image T he RF pulses are an excitation (90 ) and a refocusing (180 ) pulse, Gslice the slice encode gradient, Gphase the phase encode gradient, Gread the frequency encode or read out gradient, FID the free induction decay, AT the acquisition time, TE the time of e cho, TR the time of repetition. For magnetic resonance spectroscopy (MRS), simple hard pulse s can be applied without any gradient techniques to allow the generation of spectra. However, if the signal to be detected comes from a small specific region surro unded by regions with similar signal, localized spectroscopy will have to be applied. In this latter technique, the same gradient technique as for MRI occurs, except that it localizes a small cube region of 2 to 5 mm3 (a voxel), and the signal acquisition allows the generation of a spectrum of similar nuclei within this selected voxel (Figure 2 3B) (81) Two commonly used localized spectroscopy pulse sequences are the point resolved spectroscopy (PRESS; (82) ) sequence and the stimulated echo acquisition mode sequence (STEAM ; (83) ). The PRESS sequence (Figure 2 5) includes t hree RF pulses (one 90 and two 180 ) applied in synchronization with the gradients to generate an echo at th e time of acquisition in the
39 volume defined by the intersection of the three selective -pulse -defined planes (voxel). The sequence is T2-w eighted due to its spin echo characteristics (long TE) and may be difficult to implement to detect molecule s with a sh ort T2. Furthermore, 180 pulses are harder to achieve and requires more power than 90 pulses. Figure 2 5. A PRESS pulse sequence program example to acquire a localized spectrum The RF pulses are an excitation (90 ) and a refocusing (180 ) pulse, Gslice the slice encode gradient, Gphase the phase encode gradient, Gread the frequency encode or read out gradient, FID the free induction decay, AT the acquisition time, TE the time of echo, TR the time of repetition. The crusher gradients are correction gradients that manipulate the phase of the unwanted signals due to non -ideal refocusing RF pulses to eliminate them. The STEAM sequence (Figure 2 6) includes three 90 pulses applied in synchronization with the gradients to generate a stimulated echo at t he time of acquisition in the volume defined by the intersection of the three selective -pulse defined planes (voxel). The sequence is not as T2-
40 w eighted as the PRESS as its effective TE is shorter than the PRESS sequence TE Since o nly 90 pulses a re use d, the power requirements for this sequence are lower than the PRESS. However, the signal sensitivity is about 50% less than the PRESS sequence due to the formation of stimulated echo Furthermore, it is m ore subject to diffusion weight ing error because the spoil er gradient s are more spaced in time than the crusher gradients are in the PRESS sequence. Figure 2 6. A STEAM pulse sequence program example to acquire a localized spectrum. The RF pulses are an excitation (90 ) pulses, Gslice the slice enco de gradient, Gphase the phase encode gradient, Gread the frequency encode or read out gradient, FID the free induction decay, AT the acquisition time, TE the time of echo, TR the time of repetition. The spoiler gradients are correction gradients that dephase non -selectively any residual transverse magnetization. The PRESS sequence will initially be given the advantage when performing localized spectroscopy because of its higher sensitivity and its use in previous studies for 1H detection
41 (5,6) The STEAM sequence will be implemented subsequently, if the results obtained using the PRESS sequence are not judged satisfactory. 2 3 2 2 Nuclei of interest There are a number of biologically relevant nuclei that are detectable with NMR. For a BAP, they include, but are not limited to, hydrogen (1H), phosphorus (31P), and fluorine (19F). Hydrogen (1H) : Hydrogen is the one of the most sensitive NMR nucle i and the most abundant in living tissues primarily due to water It is commonly detected to yield clinical images (MRI). Other important biological compounds containing 1H and in the millimolar range (such as choline and lactate ) can be detected However, to be able to observe them clearly on spectra (MRS) the large water signal has been suppressed. Water suppres sion methods have been developed (84) extensively for this purpose Understanding the biochemical significance of these detectable com pounds is essential to the interpretation of acquired spectra. Lactate signal relates to an increased rate of anaerobic glycolysis (anaerobic energy metabolism) indicating the presence of hypoxia (85) or a problem with the tricarboxylic acid (TCA) cycle Choline is a vital amine molecule, which pa rticipates in cell -signaling, serves as an acetylcholine precursor, and adds structural integrity to cell membranes (Figure 2 7 ). The choline NMR signal, also referred to as total choline (TCho) signal includes the signal from the following aqueous compo unds: phosphocholine (PCho, a precursor of membrane phospholipids), glycerophosphocholine (GPCho, a product of the degradation of membrane phospholipids) and free choline (Cho). Variations in the signal intensity represent an increase of membrane phosphol ipid biosynthesis and serve s as an active marker for cellular proliferation in tumors (86,87) For a BAP, TCho intensity changes have been shown to be linked to oxygen concentrati on (88) and proportional to the change in cell viability (5,6)
42 Figure 2 7 Molecular stru cture of Choline. Phosphorus (31P ): P hosphorus NMR has been used extensively to investigate the bioenergetic status of cells, phospholipid metabolism (89) and intracellular pH (90) Despite its low sensitivity (only 6.6% of 1H sensitivity) p hosphocreatine (PCr), adenosine triphosphate (ATP Figure 2 8 ) and inorganic phosphate (Pi) are clearly observable on 31P spectra. For a BAP 31P NMR (ATP detection) has been utilized in vitro to investigate the link between cell bioenergetics and insulin secretion following step changes in glucose or oxygen concentration in TC3 cells (91,92) Figure 2 8 Molecule structure of ATP.
43 Fluorin e (19F) : Detection of 19F by NMR can be biological ly relevan t, because following excitation, the fluorine spin lattice relaxation rate characteristic (1/T1) increases linearly with the concentration of oxygen in the solution. This is due to the dipolar (paramagnetic) interactions between oxygen and fluorine molecules (93) Furthermore, t he high sensitivity of fluorine is advant ageous (83% of that of 1H) Even though fluorine is not commonly found in biological samples, the addition of a small amount of perfluorocarbons to a culture media can allow for determination of the oxygen concentration in a cell suspension For a BAP, 19F NMR is used to monitor the oxygen tension within a construct in vitro (94,95) and in vivo (96) Different perfluorocarbons (PFC) can be used. The two most frequently encountered which will be considered in the presented work are perfluorotributylamine (PFTBA, Figure 2 9 A ) and perfluoro 15 crow n 5 ether (PFCE, Figure 2 9 B). Their entrapment is usually simple and the size of their molecules prevent s their clearance from within the gel material used for the BAP keep ing the NMR detection sensitivity constant over time Figure 2 9 Molecule str ucture of the most commonly used perfluorocarbons. A) PFTBA ; and B) PFCE.
44 2 3 2 3 N uclear magnetic resonance sensitivity The main issue of the NMR techniques is their relative insensitivity (e.g., requiring millimolar concentration ranges for detection), often ch aracterized by their signal to noise ratio (SNR). For a fixed volume/number of cells, the SNR limitation can be improved upon by using higher magnetic field strengths (97) This works because the net magnetization increases with increased magnetic field strength (i.e., more nuclei align with the magnetic field, the lower energy state) (7). Another method to improve the sensitivity of the system is to enhance the RF coils. Modifying the RF coil design by matching the si ze, shape, and proximity to the structure of interest can enhance the sensitivity of the MR detection as well as the selectivity of the MR signal (8). 2 3 2 4 Radiofreq u ency coils Th e sensitivity enhancement can be seen in the evolution from simple surface coils to quadrature surface coil combinations (98) and/or phased arrays of surface coils (99) T he development of implanted coils can further increase the MR measurement sensitivity from deep specific internal structures of a relatively undisturbed system (100,101) In addition, t his coil arrangement allows for better spatial localization and will be considered in the presented work Implantable coils : Implantable or implanted coils were first developed to produce high quality in vivo phosphorus spectra (9,10) These early systems included only one coil with wires going through the skin to connect the coil with the electronic circuit and amplifier. They were also often only implanted temporarily. Recently, implantable coil sys tems have been modified to eliminate the external connection s (11 13) The modification consists of creating a s ingle unit using two coils; one coil is implanted in the body (implantable coil) while the other one is positioned extracorporeally (surface coil) The surface coil excites the implantable coil and receives its response signal wirelessly. The two coils a re inductively -coupled. Many types of
45 coils can be used in this system such as simple wire loop, loop-gap resonators (102) solenoid (8), or bird -cage resonators (103) These inductively-coupled systems represent an advantage over a single implanted coil as no cable is going through the skin to connect the implanted coil to the power source and a mplifiers (12,14,1921) The risk of infection is largely decreased, thus increasing the time an animal may be safely monitored without addition surgery Receive -only coils : In order to get a more uniform excitation pattern within the sample, large surface co ils or volume coils can be used. A smaller coil closer to the area of interest, such as the coil system described in the Implantable Coils section can then detect the FID signal with a higher sensitivity. However, the two coil sets would interact with e ach other via mutual inductance, causing the acquisition of erroneous data. One way to isolate the coils from each other is to make the RF magnetic field they produce orthogonal to each other. Nonetheless, to obtain a proper decoupling between the coils requires a high degree of accuracy in the placement of one coil compared to the other and is often not pursued, nor will it be here. The second possibility is to insert a decoupling circuit into the coil circuit (104106) This decoupling circuit added in parallel with a coil tuning capacitor will introduce a high impedance in the coil circuit, when turned on, preventing the current to flow. This decoupling can be passive or active. Unlike passive decoupling, active decoupling requires an external current source to be activate d. This distinction is why the active decoupling (104,106) is used often in larger excitation coils when the amount of current supplied to activate the diodes is an issue. P assive decoupling is used in sm aller receiver coils (105) as the larg er excitation coils usually produce enough power to active the diodes of the decoupling circuit A ctive decoupling was quickly ruled out for the system explored here as the inductive coupling between the two coils of the system was designed to remove any external connections from the implantable coil. Consequently, passive decoupling
46 will be investigated to make the inductively-coupled implantable coil system receive -only when using a birdcage volume coil for excitation (14) Multiple -frequency coils : Different coil systems can be developed to detect the different nuclei of interest in this project. However, since only one coil can be implanted per construct, coils that are tuned to multiple frequencies need to be considered. Th ese type s of coil s will be investigate d here to a llow for the simultaneous assessment of differ ent physiologically relevant parameters within a single construct and amenable direct cross -correlation studies between the different cell metabolic activities and functions. Multiple -frequency coils have been developed for over 30 years. These coils are mainly based on tank, or trap, circuits (107) and transformer coupled circuits (108) A large number of these coils can be found in solid -state NMR, essentially triple and quadruple frequency microcoils (109111) For in vivo studies, these multiple -frequency coils are generally birdcage volume coils (112) or surface coils (113,114) essentially double -tuned 1H and 31P, due to the importance of the biological compounds observed at these frequencies (112,113,1151 17) 1H detect ion is also often not optimized and only used for shimming purposes (114) However, every combination of two or more nuclei is theoretically possible (107,118) A few triple or more resonance coil s have been developed for in vivo application s (119,120) Gonen et al. utilized a surface coil with multiple tank circuit (1H -31P ) and another coil to allow 1H decoupling (119) while Eleff et al used tank and capacitively -coupled tank circuit coils O ne inductively -coupled double -frequency system was developed and implanted for in vivo studies (121) However, only the implant ed coil was double resonance T he surface coil was a system of two i ndependent nonresonating coils Furthermore, the implantable coil had accessible variable components and its distance to the surface coils could be changed as well.
47 These options are impossible to imple ment here as the system geometry is fixed b y the anatomy of the implantable coil recipient and the implantable coil is inaccessible by experiment al setting Consequently, no fully multiple -frequency inductively coupled coil system has yet been reported in the literature. Since t he efficiency of tank and transformer -coupled circuits to generate multiple -frequency coils is relatively similar (113) both methods will be investigated. 2 3 2 5 Coil coating Any in vitro and in vivo testing of an inductively -coupled coil system will require the development step of isolatin g the implanted coil from the host. The i mplantable coil is made of non -biocompatible material which can potentially trigger an immune response from the host. Moreover, the strong coupl ing the coil can have with the body will s ubstantially decreas e the performance of the coil as biological tissues and fluids are highly conductiv e. Consequently, coating any coil to be implanted is necessary First of all, this coating needs to be biocompatible not to trigger a chronic immune response from the host that would adver s ely affect the hosts health and prevent the implanted coil from performing optimally. The coating also needs to insulate the coil from the tissues to keep the coupling between the m as low as possible and the performances of co il optimal. Therefore the coating material must have a low dielectric constant to eliminate the electric field storage and minimize the disturbance of the magnetic field generated by the coated coil and a low conductivity to minimize the electrical losses in the sample If the dielectric constant is not low, the introduction of coupling capacitance between the coil, the coating, and the body will create a resonant frequency shift of the implantable coil. If the conductivity is not low, the loading of th e coil will decrease the performances of the coil characterized by its quality factor (Q) For the BAP, the coating will also need to be soft and flexible as it will surround the construct and will be directly in contact with body tissue when implanted i n vivo
48 A broad range of materials has been employed to coat inductively -coupled coils. Polyethylene tubing was used to coat the wires that connected the first implanted coil to its tuning and matching network outside the animal body (9,10) as well a s subsequent studies done by Schmidt et al. (122,123) However, Plastidip a solvated polymer resin was likely the most extensive ly used material (11,124127) because of its flexible coating and sealing properties. Other materials that have been used include polyurethane (12) silicone rubber (21,128) epoxy (polyepoxide) and fiberglass (129) PARAfilm (13,130) Gortex (122,131) and Parylene C (132) Most of the se polymers are commercially available for non -medical applications. T hey were employed based on their flexible properties and ease of manipulation and handling rather than on their biocom patibility. Unfortunately, the methods utilized to coat coil s have never been fully described nor have biocompatibility studies been reported, although some coils were tested after retrieval for fibrosis growth No intense immune response of the animals has ever been reported, even for coils implanted for extended periods of time (133) More recently, Silver et al. used polymethylmethacrylate as a biocompatible polymer (Figure 2 5) to coat implantable coils for spinal cord injur y studies (14) Polymethylmethacrylate : Polymethylmethacrylate (PMMA) has been used extensively in intraocu lar lenses and orthopedic devices over 40 years for its ease of handling, biocompatibility, and stability over 10 years in vivo (134) It has favorable dielectric properties for the purpose of implantation, a low diel ectric constant of 2.6 and a low conductivity of 1016S The preparation of the coating material and the coating itself were described as rather simple by the user (14) PMMA (Figure 2 10) forms from the mixing of a powder base and a solution of curing agent. It is easy to pour in any mold since the base -curing agent mi xture stays fluid for several minutes
49 before drying at room temperature in a couple of hours. The use of this coating material will be explored in the presented work as well Figure 2 10. Molecular structure of PMMA. Polystyrene : Polystyrene (PS Figu re 2 11) was found to have slightly better dielectric properties than PMMA: a dielectric constant of 2.6 and a conductivity of 1020 S. These properties have encouraged its testing as a coating material for implantable coils. Figure 2 11. Molecular st ructure of PS. It is widely used for electronic applications, such as capacitor dielectric and high frequency circuit insulation (135) PS Petri dishes and multiwell plates are also w idely used for cell culture. A dip coating procedure seems an appropriate method to use with this material for coating implantable coils (136) and will also be considered in the presented work Polydimethylsiloxane : Polydimethylsiloxane (PDMS) is a biocompatible polymer also noted for its easy handling (137) This silicon based organic polymer has been widely used in
50 biomedical applications for over 50 years due to its biocompatibility and chemical stability (Fig ure 2 1 2 ). Figure 2 1 2 Molecular struct ure of PDMS. PDMS elastomer can be found in catheters, drainage tubing, and insulation for pacemakers and membrane oxygenators, a s well as ear and nose implants (137) The biomedical grade PDMS has a dielectric constant of 3 and a conductivity of 5x1016 S. Its preparation is similar to PMMA except that both the base and curing agent are viscous solutions. Some research groups (15,1719) have already used this material to coat their implantable coils with multiple layer s and so it will be considered here. Polytetrafluoroethylene: The material with the best dielectric properties possible to be used as a coating material is polytetrafluoroethylene (PTFE, Fig ure 2 1 3 ). It is chemically inert and has low permeability to liq uids, gases, moisture, and organic vapors. Figure 2 1 3 Molecular structure of PTFE. This material has been used in vascular graft and tendon replacement for many years due to its biocompatibility, stability, and non biodegradation (138) Its dielectric constant as low as
51 1.94 and i ts conductivity is as low as 1019 S and. Nonetheless, Teflon has a low bonding affinity that may compli cate the coating procedure; hence a surface treatment to the coil is required before coating. Moreover, Teflons glass transition temperature of 160 C necessitates adaptation of the materials used for coil construction since heating above this temperature is necessary to generate the coating However, the outstanding dielectric properties of PTFE made its use as a coating material worthwhile investigating in the presented work 2 4 Summary In order to better accomplish the non -invasive i n vitro and in vivo mo nitoring of bioartificial pancreas using NMR imaging and spectroscopy techniques, this dissertation describes the work to develop i nductively -coupled implantable coil s ystems. Different designs of these implantable coil systems are considered (loop-gap re sonator and solenoid) and their sensitivities when used with NMR are characterized. Initially, t he inductively -coupled systems were single -frequency to assess the ir feasibility (Chapter 3). M ore complex systems such as receive-only (Chapter 4) and mult iple frequency systems (Chapter 5), wer e developed later to further improve the NMR method sensitivity and allow the acquisition of metabolic activity information from MRS data Passive de coupling was explored to render the coupled -coil systems receive -on ly (Chapter 4), whereas tank circuit and capacitively overcoupling was used to render the coupled -coil systems multiple -frequency (Chapter 5). Coating materials are also an important component of the system, and the studies to determine the best material (PS, PTFE, PMMA, and PDMSe) with which to coat the implantable coil (in terms of biocompatibility and dielectric properties) and entrap the insulin -secreting cells are considered in Chapter 3 T he final construct must also be able to house and maintain in sulin -secreting cells Different c onstructs we re generated via micro or/and macroencapsulation using agarose gel and alginate foam and
52 gel (Chapter 3). Their insertion in the middle of the implantable coil was tested along with their ability to hold cel ls and allow them to grow in Chapter 6. TC -tet cells were selected as the insulin -producing cell model as they a re easily available easy to culture, and their growth could be regulated if needed. T hese studies outline the many aspects required for the development of the highest sensitive NMR imaging and spectroscopy method to monitor a bioartificial pancreas and lead toward the development of a functional implantable bioartificial pancreatic construct which can be non -invasively monitored.
53 CHAPTER 3 DEVELOPMENT OF AN INDUCTIVELY COUPLED RF COIL SYSTEM FOR IMAGING AND SPECTROSCOPIC ANALYSIS OF AN IMPLANTABLE BIOARTIFICIAL CONSTRUCT AT 11.1T1 3 1 Introduction Developing a method to monitor tissue -engineered constructs non-invasively is critical for the optimization of construct design and for assessing therapeutic efficacy. NMR is a powerful technique for this purpose as it can be used to obtain both images and spectroscopic data. But the inherent sensitivity of NMR methods limits the observation of a bioartificia l construct with current RF coil technol ogy (surface coil) In this study, this limitation was addressed through the development of an inductively -coupled, implanted coil system, demonstrate its use at high field (11.1 T), and investigate the use and requ irement of this coil system for monitoring a bioartificial construct in vitro and in vivo The results establish that large gains in signal to -noise can be obtained with this inductively-coupled, implanted coil system over that which can be obtained with a surface coil. This coil system provides a means to analyze the structure and function of implanted bioartificial organs quantitatively. 3 2 Background One approach being explored to cure diabetes is to implant insulin-secreting cells within a tissue -enginee red (bioartificial) pancreatic construct (1,29,33,35) Using NMR to non -inv asively monitor an implanted pancreatic construct can provide correlations between construct function and physiologic effects post implantation (2 4) NMR also offers the possibility of assessing changes in construct fu nction towards developing early markers of construct failure in advance of end -point diabetic effects, e.g., hyperglycemia. 1 Modification of paper submitted for peer review, 2009.
54 Stabler et al. developed an NMR based method to non invasively assess the viability of a bioartificial pancreatic construct (5, 6) Their method, correlating the 1H choline signal with cellular viability, was accomplished both in vitro and in vivo on macroconstructs using a 4.7 T NMR instrument and a surface coil (SC) as the RF coil for 1H NMR excitation and detection. However, a shortcoming of this measurement design is the threshold sensitivity, which at best is limited to a cellular density of seven million viable cells/ml of construct under ideal in vitro conditions. Because of the desire to study implanted constructs in viv o and also to study lower cellular densities, improving the sensitivity of the NMR measurement is critical. Two changes can improve the sensitivity of the NMR measurements and decrease the lower limit of detectable cells: (i) acquire the NMR data at a hig her field; and (ii) improve the sensitivity of the RF coil used to measure NMR. Increasing the static magnetic field strength from 4.7T to 11.1T would improve the sensitivity of the measurements by a factor of 2.4, since the sensitivity, characterized by the signal to -noise ratio (SNR), increases linearly with magnetic field strength in lossy samples, such as tissue (7). Furthermore, t he development of an implant able coil (IC) that not only matches the size and shape of the construct, but is also inductively-coupled to a n SC would also increase the sensitivity of measuring specific in ternal structures yet leave the subject relatively undisturbed (11 13,100,101) Additionally, t hese i nductively-coupled, implantable coil systems have an advantage over implanted coil s since no cable passes through the skin to connect the implanted coil to the power source and amplifiers (12,14,1921) The refore, risk of infection is decreased, increasing th e period over which an animal may be safely monitored. This report investigates the use of inductively-coupled implantable coil systems to monitor a n implanted pancreatic construct The NMR sensitivity of these systems was then compared to
55 the sensitivity of an SC at a high field (11.1 T). To assess NMR sensitivity improvement of the inductively-coupled system under development compared to the SC used at 4.7T in the previous studies (5,6) an SC was also tested at two magnetic field strengths (4.7 a nd 11.1 T). Lastly, this report discusses important technical aspects toward generating inductively -coupled implantable coil systems, and issues to consider when applying these RF coils in conjunction with implanted tissue -engineered constructs 3 3 M aterials and Methods 3 3 1 RF Coil S ystem D evelopment Four different inductively -coupled implantable coil systems using two different implantable coils and two different configurations were constructed, optimized, and tested for NMR imaging and spectroscopy at 11.1 T, t o study the NMR sensitivity dependence of RF coil design. These systems consist ed of an IC inductively -coupled to a n SC, functioning both transmitter and receiver for 1H NMR (470 MHz at 11.1 T). The IC surrounds a bioartificial c onstruct adapted from St abler et al. (5,6) allow ing the coil to be near the monitored cells (Fig ure 3 1 ). Figure 3 1. Implantable coil bioartificial pancreas macroconstruct assembly diagram. A) Transverse view and B) coronal view.
56 T wo types of implantable coil a lo op-gap resonator and a solenoid, with identical inductor cross -section (1.2 cm) and thickness (2 mm) were built and tested to determine the design with the highest sensitivity. These designs were considered because: (1) the solenoid design may provide a higher quality factor (139) ; (2) the loop -gap resonator is perhaps easier to manufacture and may perform better at higher frequency (102) A rod was used to build these implantable coils and keep their cross -section constant. A spacer was used afterwards to check their flatness and thickness. T he inductive part of th e IC (Li on Figure 3 2 A ) was constructed with 20-AWG heavy a rmored poly thermaleze magnet copper wire (Belden, St. Louis MO ) for the solenoid and with a 2 mm-wide 202m -thick copper foil (McMaster Carr, Atlanta, GA) for the loop -gap resonator. A single f ixed tuning capacitor (American Technical Ceramics, Hartford, CT, CTi on Fig ure 3 2 A ) was used to provide the desired resona nce for both coil types Th is capacitor was soldered between the wire ends after these ends were twisted one half -turn ( approximate ly 2 mm total length ). This construction makes the coil more stable and displace s the tuning c apacitor away fr om the inductor The loop-gap resonator capacitor was soldered directly between the copper foil ends. The SC were constructed with a single 2 cm diameter circular loop of 2 mm wide, 35 m thick copper tape ( 3M, Saint Paul, MN ; Ls on Figure 3 2 B) placed on a piece of 5 mm thick Plexiglas. The capacitive portion of the SC includes two variable matching capacitors (Voltronics Corporation, Denville, NJ 1 15 pf ; CM s 1 V and CM s 2V on Figure 3 2 B), one fixed tuning capacitor (American Technical Ceramics, Hartford, CT ; CTs on Figure 3 2 B), and one variable tuning capacitor (CT sV on Figure 3 2 B 1 15 pf). The fixed capacitor was placed on the inductive loop, and three variable capacitors were soldered onto a circuit board to provide
57 variable tuning and matching of the system. A coaxial cable connected the coil to the source and preamplifier. A cable trap was added to the shield of the coaxial cable to block shield currents that contribute to parasitic coupling when working at high frequency (140) This SC of the inductively-coupled coil system was also used as an isolated SC since it could be tuned and matched optimally at 470 .75 MHz. Furthermore, to compare these results with previous measurements performed at 4.7 T with an SC (5,6) a separate SC was built with the same geometry as the SC of the implantable system presented above, then optimized to resonate at 200 MHz (1H frequency at 4.7 T). Figure 3 2. Circuit diagram for an inductively -coupled implantable coil sy stem for 1H detection. A) Implantable coil; and B) surface coil. Inductors (L) and capacitors (C) are shown with subscripts denoting implantable (i) or surface (s) coil Capacitors are also designated as tuning ( CT) or matching ( CM). M indicates the mu tual inductance or inductive coupling between the two coils of the system and V variable A cable trap equivalent, including an inductor and a capacitor, is also shown in the diagram. 3 3 2 Coat ing T echnique s T he implantable coil required isolation from the hos t to allow it to function when testing the inductively-coupled coil system in vitro and in vivo Four different materials were chosen and tested to coat the implantable coil: 1) p olystyrene (PS); 2) polytetrafluoroethylene (PTFE);
58 3) p olymethylmethacrylat e (PMMA), and 4) p olydimethylsiloxane elastomer (PDMSe). These materials were chosen for their biocompatibility and low dielectric properties (see S ection 184.108.40.206). The coating uniformity, seal and reproducibility were the first parameters checked. T he effects of the coating s and their thickness on the coils were tested afterwards by determining changes in the coil resonant frequency and quality factor (Q) both in air and in a gel phantom (139) The material having the easiest handling and the least influence on the coil performances was selected for the in vitro and in vivo studies. 3 3 2 1 Poly s tyrene Polystyrene (PS, Resirene, Mexico) was formed by first heating pellets at 105 C (Tg for PS) for 20 minutes in a dry oven to evaporate any water and avoid bubbles during the coil coating. PS pellets (1.56 g) were then placed in a closed container of chloroform (20 mL) and stirred at room temperature until completely dissolved. This solution was a 5% (w/ v) PS solution (136) The implantable coil was dipped into the solution and hung for 30 minutes at room temperature to allow the coating to dry. The operation was repeated up to 7 times. Each time, the coil was hung at a different locatio n around the main coil inductor to even the overall coating. 3 3 2 2 Polytetrafluoroethylene A solution of polytetrafluoroethylene (PTFE) polymer in solvent fluorinert FC 75 (Teflon AF 1600, DuPont Fluoroproducts, Wilmington, DE ), 6% by weight, was used to coat t he implantable coil. To increase the bonding properties of PTFE, the coil was first sanded to increase surface roughness; then wetted with an acid (solder flux in this case) and flamed to oxidize the coil surface. Traces of the wetting agent were removed by cleaning, rinsing, and blow drying the coil. T he coil was then dipped in the PTFE solution for about 1 minute, as described by Regan et al. (141) and withdrawn from the solution very slowly to avoid bubble
59 form ation The coil was hung and allowed to dry in the oven at room temperature for 5 to 10 mins. The temperature was then slowly increased up to 165 C over a one -hour interval which ma d e the polymer chains more mobile and enhance d solvent evaporation leadin g to fewer defects in the dip coating. An additional 15 minutes at 165 C was attempted to improve the quality of the coating and reduce the presence of air bubbles. This process was repeated up to 3 times with the coil hung from different locations aroun d its m ain induct or to produce an even coating. Another alternative was to hea t up to 20 mL of PTFE solution to 115 C (solvent evaporation temperature + 5 C) in a small beaker. With a 6% PTFE solution, a 1.5 cm height solution left a 2 3 mm film at the bo ttom of the beaker. The process was repeated after a coil was placed on the first film. To make sure that the temperature would not affect the coil performances, the capacitors were upgraded to high temperature capacitors able to withstand temperatures u p to 310 C instead of the regular 125 C. The solder used to connect the capacitor to the inductive loop was also upgraded from a 60%Sn / 40%Pb with a melting point of 183190 C to a 5%Sn / 93.5%Pb / 1.5%Ag, whose melting temperature is 296302 C. 3 3 2 3 Polym ethylmethacrylate Polymethylmethacrylate (Cranioplastic, Plastics one, Inc., Roanoke, VA) was formed by mixing a powder base and a solution of curing agent. The powder of PMMA particles includes diethyl phthalate a plasticizer to render the PMMA less brittle and softer The curing agent is a thin sol ution of methyl methacrylate monomer which also contains an initiator (N,N dimethyl -p toluidine) that promote d PMMA chain formation at room temperature The ratio by weight of base to curing agent was set at 1:10. This allowed the polymer to acquire the desired consistency to hold its shape, yet maintain its flexibility. The mixture was then poured in a mold to create a
60 1mm thick layer which was let to dry at room temperature overnight. The implantable coil was then added on top of the layer and another PMMA mixture was poured to cover the coil and add another 1 mm on top of the coil. The coated coil was easily removed from the mold 24 hours after the second casting. 3 3 2 4 P oly di methylsiloxane Poly di methylsiloxa ne (Medical Grade Silicone Elastomer, Factor II, Lakeside, AZ) preparation was similar to the PMMA one except that both the base and curing agent we re viscous solutions and the mixing proportions were a 10:1 base to curing agent ratio D imethylvinyl terminated dimethyl siloxane made the base solution including r einforcing filler particles of trimethylated silica to fill the PDMS matrix increase its tensile strength and make it mechanically more stable. A platinum catalyst was also present in the base so lution to make the crosslinking possible. D imethylvinyl terminated dimethyl siloxane and dimethyl,methylhydrogen siloxane (crosslinker) composed t he c uring agent. The mixing of these solutions require d the degassing of the mixture for at least 20 minutes under vacuum (95 102 kPa) since both solutions we re very viscous and many bubbles get trapped in during the mixing. The coils were either dip coated repeatedly up to 3 times such as done with PTFE or they were casted. The casting was done in two steps to insure a uniform and sealed coating with the coils centered in the middle of the coat A first PDMS elastomer layer w as generated by curing a PDMS base for 24 hours at room temperature between glass plate s carefully spaced to give the desired coating thi ckness Once the PDMS was cured, the top glass plate was removed and several implantable coil s w ere placed on top of this first PDMS layer carefully spaced from one another. A second PDMS layer was then poured on top of the coils. In order to obtain a u niform coating thickness around the coils, a glass plate was placed above the cur ing second PDMS layer
61 carefully spaced from the bottom glass plate. As PDMS is a crosslinked polymer, the two layers bond to each other when curing, therefore embedding the c oils in a sealed coating. The curing process could be accelerated by heating the layers for about 1 hour at 50 C. The coated coils were then punched out of the PDMS slab using hollow punches A smaller punch removed the extra PDMS in the center of the c oil to allow the insertion of the bioartificial construct. 3 3 3 RF Coil Testing When surface and implantable coils are in proper orientation and relative proximity, they inductively couple. When the two coils are far apart, their mutual inductance is low, they are loosely coupled. Their coupling or mutual inductance increases as the coils become closer to each other. For surface and implantable coils tuned to the same resonant frequency, critical coupling occurs when the IC resistance coupled into the SC is e qual to the surface coil resistance (142) If the coupling strengthens the coils overcouple and the resonan t frequency of the coupled system splits as two current modes appear: a co rotating (+) and a counter rotating ( -) mode (Figure 3 3 ; (143) ). Three different configurations of the inductively-coupled coil system were then tested: 1) a loosely-coupled system where the SC was not resonant and the IC resonates at 470 .75 MHz; 2) an overcoupled system where both surface and implantable coils resonate at the same frequency in the (+) mode); and 3) an overcoupled system where both surface and implantable coils resonate at the same frequency in the ( ) mode. In the (+) mode, each coil was tuned and matched individually to a higher frequency (490.1 MHz), so that the lower resonant frequ ency of the overcoupled system occurred at the desired 1H NMR resonance of 470.75 MHz I n the ( ) mode, each coil was individually tuned and matched to a lower frequency (450.75 MHz) so that the higher resonan t frequency of the overcoupled system occurre d at 470.75 MHz (Figure 3 3). T he capacitors CM1 V, CM2 V, and CT sV (Fig ure 3 2 ) of the
62 surface coil were adjusted to achieve the match of the whole system to 50 at 470.75 MHz. The systems and configurations were simulated first with antenna analysis sof tware (Graphical Numerical Electromagnetics Code, Nittany, Inc. Riverton, UT), then constructed and optimized on the bench befo re being tested in the magnet. Figure 3 3. Voltage standing wave ratio (VSWR) of a surface coil (SC), an implantable coil (IC), and an inductively over coupled system in its different modes A) B oth the SC and IC resonate at 470 MHz initially creating two o ffresonance modes; B) both the SC and IC resonate at 490.1 MHz initially allowing the (+) mode to be at 470.75 MHz ; and C) both the SC and IC resonate at 450.75 MHz initially allowing the ( ) mode to be at 470.75 MHz
63 The characteristics of each coil taken individually and of the whole system (resonant frequency and Q) w ere determined after construction with a vector network analyzer (VNA Hewlett Packard 8752C, Santa Rosa, CA) with and without load and every time the system was tested in the magnet (144) The Q was measured by finding the 3 dB points (or bandwidth) from the VNA return loss plot for the SC and for the inductively-coupled coil system hooked directly to the network analyzer via its coax ial cable (Fig ure 3 4A ), and from the VNA transmission plot for the IC (Figure 3 4B). The implantable coil, which has no direct coaxial c onnection, was loosely coupled with two probes connected to the reflection and transmission ports of the netwo rk analyzer. As the implantable coil placed in between the two probes absorbed energy, the absorption response was viewed on the transmission plot of the network analyzer allowing Q determination. The VNA calibration was saved for a specific center frequency and bandwidth and checked periodically using a 50 load, a short, and an open for the reflection line, and a through cable for the transmission line. Figure 3 4. Vector network analyzer outputs used for RF coils Q measurements. A) Return loss plot; and B) transmission plot. Marker 0 was placed at the resonan t frequency of the coil or system being tested and Marker 1 and 2 at the 3 dB points around the resonant frequency.
64 Additionally, the voltage across the implantable coil was also determined w hen the coil was coupled with the surface coil to insure the power deposited within the construct was appropriate and would not deteriorate the construct viability and functionality (145) This voltage was measured directly across the capacitor of implantable coil via a half wave coaxial cable connected to a high impedance port of a digital oscilloscope Tektronix DPO 4104. The surface coil was powered by the RF pulses from the spin echo (SE) sequence used to image the construct and described in S ection 3.3.7. The implantable coil was place d a round a water sample as described in S ection 3.3.4. 3 3 4 Samples A sample of distilled water placed in a 1 0 mm -diameter thin -wall glass tube (Wilmad Labglass, Buena, NJ) was used to test the coupled -coil systems (with coated and uncoated IC) on the bench and in the magnet. This water phantom was considered to represent the unloaded condition because of its small size compared to the SC and the nature of the sample. To test performances under loaded conditions (similar to in vivo conditions), the implantable co ated coil was placed in a mouse abdomen -like gel phantom (146148) comprised of a viscous 6.7% (w/w) polysaccharide gel (TX 151; Oil Center Research, International LLC, Lafayette, L A). The permittivity of this gel phantom was determined to be at 64.2 1.13 ( 1.77%) and its conductivity at 1.13 0.01 ( 1.29%) S/m. I nductive coupl ing was accomplished by placing t he water sample or phantom (with IC) on top of the SC. The IC were tested when placed around the water sample no more than a 1 cm away from the SC, and when placed in the gel phantom 0. 5 cm and 1 cm away from the SC.
65 3 3 5 Construct D evelopment Three different constructs were also developed in this study to select the most appr opriate one to function as a BAP when surrounded by an implantable coil. The constructs tested were : 1) agarose gel disks (5,6) ; 2) alginate sponge or foam disks (149,150) ; and 3) alginate beads (33,37,55) The diameter of the construct could not exceed 1 cm in order to fit within the middle of the coated implantable coil (Fig ure 3 1 ) and its thickness di d not exce ed 500 m to minimize the diffusion limitations. 3 3 5 1 Agarose A 2% agarose gel solution was prepared by suspending dry agarose powder (SeaPlaque Low Melting Agarose, Cambrex Bioproducts, East Rutherford, NJ) in Phosphate Buffer Saline (PBS Invitrogen, Carlsbad, CA ) without Calcium and Magnesium and boiling the covered mixture until a clear solution forms (about 10 minutes). Drops of this solution were then poured in the middle of the coated coil and allowed to cool down to room temperature to form a rigid agarose gel slab. The coil served as the mold. 3 3 5 2 Alginate foam Alginate foam was purchased (Novomatrix, Trondheim, Norway) Circular, 1 -cm diameter pieces were punched out from this foam and placed in the middle of the coated coil. Non absorbable su rgical sutures were used to hold the alginate foam construct in place in the middle of the coil. 3 3 5 3 Alginate beads A 2% (w/v) sodium alginate solution was prepared by dissolving an alginate powder with 62% / 38% mannuronic/guluronic content (Novomatrix, Trond heim, Norway) into a saline solution (0.85% NaCl) overnight Cell -free a lginate beads were generated as described by Simpson et al. (151) by running the alginate solution through a s yringe at a constant flow rate
66 and dropping it in a cationic cross linking solution with an electrostatic bead generator (Nisco, Zurich, Switzerland) to set the bead diameter between 400 and 600 m To keep these beads together in the middle of the implantable coil, a biocompatible polyetheretherketone (PEEK) 300m mesh (Small Parts, Inc., Miami Lakes, FL) was added on the top and bottom of the ring formed by the c oated coil to create the construct cavity. Aliquots of freshly made beads (~0.3 ml each) we re transferred into the implantable coil -construct assembly us ing a syringe and 16 G needle. The three implantable coil -construct assemblies were initially placed into Hanks Buffered Saline Solution (HBSS) for up to 2 weeks to determine the most appropria te construct to be used for in vitro and in vivo studies. However, prior to the BAP insertion in the middle of the implantable coil for in vitro and in vivo studies, t he coated coil had to be prepared in a two -step process. It was first acid washed and r insed to remove proteins left on the coil during manufacturing and reduce the immune response to the chosen device when placed in a mouse abdomen It was afterwards autoclaved to become sterile and able to receiv e the pancreatic construct piece (agarose s lab, alginate foam, or alginate beads) For in vivo studies, the selected coil -construct assembly was placed into sterile HBSS until implantation HBSS was chosen as a culture media substitute. It is the basis of complete culture media except it does not contain the amino acids and vitamins a complete media has, which would trigger an immune response in vivo This solution also contained antibiotics (100 U/ml penicillin and 100 ng/ml streptomycin), anti inflammatory and immunosuppressant (100 nM dexametha sone). 3 3 6 Animal P reparation and H andling Experiments with female C3H/HeN mice, weighing 20 30 g, were conducted according to a protocol approved by the University of Florida Institutional Animal Care and Use Committee. NMR measurements and surgeries were co mpleted under general anesthesia. Anesthesia was
67 induced by inhalation of 2% isoflurane in oxygen and maintained for the duration of the surgery or NMR measurement by ventilating a mixture of 1% isoflurane in oxygen via a nose cone. Under deep anesthesia, the sterile construct was implanted into the peritoneal cavity via a small (2 cm) midline celiotomy. The animal was placed prone on the surface coil/cradle apparatus for NMR measurements afterwards. The construct was either surrounded by an IC, to crea te the inductively-coupled coil system when placed on the coil/cradle apparatus, or a coil -free construct studied with an SC to compare an inductively -coupled implantable coil system to an SC. The animal respiration and skin temperature was monitored usin g a monitoring and gating system (Small Animal Instruments, Inc., Stony Brook, NY). Its respiration was maintained between 20 and 30 breaths/min and its skin temperature between 24 and 26 C. Upon completion of NMR measurements, animals were euthanized a nd the coil construct assembly or coil -free construct were retrieved. 3 3 7 NMR Measurements NMR measurements were performed on an 11.1 T 40 cm clear horizontal bore Magnex magnet equipped with a Bruker Avance console and on a 4.7 T 33 cm horizontal bore Oxford Magnet equipped with a Bruker Biospin console (Bruker Instruments, Billerica, MA). Data acquisition and processing were done using Bruker Paravision software. The SNR was determined for each coil system with custom image analysis software written in IDL (ITT Visual Information Solutions, Boulder, CO). For coil system comparisons with phantoms, 1H images were acquired using a spin -echo (SE) pulse sequence with a repetition time (TR) of 1000 ms, an echo time (TE) of 10 ms, 1 -mm slice thickness, 1 average, 4x4cm2 field -of -view (FOV) and a 256x256 matrix. Coating s ignal was suppressed when required using a spectrally selective saturation pulse centered on the coating signal For in vivo coil system comparisons, 1H images
68 were acquired using a SE pulse sequ ence with the parameters given above, except that the TR was lengthened to account for the animal breathing and reducing its artifacts To reduce breathing artifact for the in vivo studies, other parameters were varied as well, such as number of averages and number of averaged experiments. Conventional r espiratory gating was also tested To compare results obtained at 11.1 T and 4.7 T, differences in relaxation times of the phantom water were taken into account. De Graaf et al. have shown that the longitudinal relaxation time (T1) of water and metabolites increases whereas their transverse relaxation time (T2) decreases with magnetic field strength (152) Therefore, a long TR and short TE were used at both magnetic fields to minimize the signal dependence of these parameters. NMR images were acquired using an SE pulse sequence with the following parameters: TR of 15, 000 ms, TE of 10 ms, 1 -m m slice, 2 averages, 4x4 cm2 FOV and a 64x64 matrix on both systems. 3 3 8 SNR M easurement and S tatistical An alysis Five different coil -systems/configurations were tested in the magnet: 1) SC; 2) coupledcoil system with an implantable solenoid in the co rotati ng (+) mode; 3) coupled-coil system with an implantable solenoid in the counter rotating ( ) mode; 4) coupled -coil system with an implantable loop -gap resonator in the co rotating (+) mode; and 5) coupled-coil system with an implantable loop -gap resonator in the counter rotating ( ) mode. The SNR of these different coil -systems/configurations was evaluated from the NMR images by choosing a signal region-of interest (ROI) at the IC position and a noise ROI outside of the sample. Saved signal and noise 770pixel ROIs were used for each SNR determination. The Henkelman method (153) was utilized to calculate the SNR. Image s in three orthogonal directions (transversal, sagittal, coronal) were acquired for each system/configuration. The number n, of i ndependently -made coil systems tested to assess the reproducibility of the system construction is indicated for every experim ent by (n) At least two set of images were acquired for each coil system to assess the
69 reproducibility of signal detection and the sensitivity of the systems. The SNR determined for each similar system/configuration in the three orthogonal direction was then averaged and normalized to the SNR of SC at 4.7 T (normalized to 1). This normalization was appropriate because the SC a t 4.7 T had the same size and shape as the SC at 11.1 T. All the results are indicated as a mean standard deviation (relative s tandard deviations are indicated in parentheses in percentage). The propagation of error was included in the normalization stati sti cs. A statistical analysis was performed using a two -sample unequal variance t -test. The r esults of this analysis were con sidered statistically significant w hen p-values were less than 0.05 3 4 Results 3 4 1 RF Coil System The loop -gap resonator and solenoid IC (Figure 3 5) as well as the SC (Fig ure 3 6 ) were successfully built and tested on the bench and in the magnet. However, the loosely -coupled coil system performed poorly on the bench because the coupled -coil system could not be tuned or matched to 50 once the IC was coupl ed to the SC under variable conditions of coil loading (16) Furthermore, the establishment of critical coupling was never achieved in this case since the size and position of the two system coils are fixed by the experiment settings. Due to those limitations, this configuration was not included in this study. Figure 3 5. Photographs of implantable coils developed as part of an inductively -coupled implantable coil system A) Loop -ga p resonator and B) Two -turn solenoid. (Photo by N.A. Volland).
7 0 Figure 3 6. Photograph of the surface coil developed as part of an inductively-coupled implantable coil system This surface coil design was also used as a single surface coil at both 470 and 200 MHz. (Photo by N.A. Volland). The characteristics of the different coils and overcoupled systems tested are summarized in Table 3 1. The implantable loop gap resonator has an average Q of 203.90 6.20 ( 3.04% ) when resonating at 491.81 3.38 (0.6 9 %) MH z an average Q of 151.8 11.6 ( 7.64%) when resonating at 451.4 2.4 (0.53%) MHZ, the implantable solenoid an average Q of 180.0 19.0 ( 10.6%) when resonating at 497.2 5.21 ( 1.05%) MHZ, and the implantable solenoid an average Q of 158.5 20.3 ( 12.81%) when resonating at 450.2 1.62 ( 0.36%) MH z. T he surface coil has a Q of 72.2 4.95 ( 6.86%) when resonating at 493.2 MHZ, a Q of 87.4 when resonating at 450.8 MHZ, an average Q of 87.0 5.67 ( 6.52%) when resonating at 470 MHZ and an average Q of 69.7 1.98 ( 2.84%) when resonating at 200 MHz. W hen resonating at 4 70.75 MH z, the inductively -coupled implantable coil system including the LG implantable coil has a Q of 96.21 3.70 ( 3.85%) in the (+) mode, a Q of 89.35 4.17 ( 4.66%) in the ( ) mode, whereas the inductively-coupled implantable coil system including the solenoid implantable coil has an average Q of 92.80 2.65 ( 2.85%) in the (+) mode and an average Q of 71.18 6.65 ( 9.35%) in the ( -) mode.
71 Table 3 1. I mplantable coil (IC), surface coil (SC) and inductively -coupled implantable coil system characteristics Resonan t frequ ency (f) and quality factor (Q) for : SC at 4.7 T (SC 4.7T) and at 11.1 T (SC), IC solenoid (IC_Sol ) and loop -gap (IC_LG ), SC (SC ), a nd coupled -coil systems (Sol and LG ) in the ( ) mode, and IC solenoid (IC_Sol+) and loop -gap (IC_LG+), SC ( SC+ ), and coupled coil systems (Sol+ and LG+) in the (+) mode. Coil # # of coils tested* f (MHz) Q IC_LG+ 7 491.81 3.38 ( 0.6 9 %) 203.90 6.20 ( 3.04% ) IC_LG 4 451.4 2.4 ( 0.53%) 151.8 11.6 ( 7.64%) IC_Sol+ 5 497.2 5.21 ( 1.05%) 180.0 19.0 ( 10.6%) IC_Sol 4 450.2 1.62 ( 0.36%) 158.5 20.3 ( 12.81%) SC+ 2 493.2 72.2 4.95 ( 6.86%) SC 1 450.8 89.3 2.69 ( 3.01% ) SC 4 470.75 87.0 5.67 ( 6.52%) SC 4.7T 1 200.0 69.7 1.98 ( 2.84%) LG+ 5 470.75 96.21 3.70 ( 3.85%) LG 4 470.75 92.80 2.65 ( 2.85%) Sol+ 5 470.75 71.18 6.65 ( 9.35%) Sol 4 470.75 89.35 4.17 ( 4.66%) All coils taken individually and all systems were tested under unloaded conditions. Even when there was only one coil built, it was tested more than once. Figure 3 7 shows the resonan t frequencies and the B1 magnetic field produced by a SC alone, an overcoupled system in the (+) mode, and an overcoupled system in the ( -) mode. All the RF coils were optimized to resonate at 470.75 MHz The magnetic field was calculated along t he axis of symmetry of the coils with the SC position set at 0 and the IC centered at 1 cm (macroconstruct location). The position of the simulation and images are aligned in the horizontal direction for ease of comparison. The (+) mode (Figure 3 7I & 3 7J ) corresponds to
72 the lower resonant frequency of the system (Fig ure 3 7C ) where the currents in the two coils rotate in the same direction and their magnetic fields are additive. The ( -) mode (Fig ure 3 7G & 3 7H ) corresponds to the higher frequency of the system (Fig ure 3 7B ) where the currents in the two coils rotate in opposite direction and their magnetic fields are subtractive resulting in a null between the two coils (14,143) The peak voltage developed across the capacitor of the implantable coil varied from less than 2 V up to 24 V when the implantable coil was coupled with the surface coil and the RF pulse power was scaled within a large range of settings (over 20 dB) beyond the experimental range to obtain the proper excitation for a SE sequence. This voltage range is much lower than the one developed across the surface coil as the coupling factor between the two coils was estimated at 0.083 (142) Furthermore, k nowing that the power requirement for an inductively coupled coil sy stem is much less than the power requirement for a single surface coil (10 dB on average) the power deposited in the mouse was revealed to be much lower than the power deposited by a surface coil used alone and no further power measurements were performed (154)
73 Figure 3 7. RF coil frequency traces, simulated B1 magnetic field magni tude, and their respective corresponding spin -echo images along the axi s of symmetry of the coils. These graphs are for a surface coil ( A, D, E, & F) ; an inductively-coupled coil system in the ( -) mode ( B, G, & H) and in the (+) mode ( C,I, & J ) at 11.1T In D through I t he SC position was set at 0 and represented by a solid line, while the IC is located at 1 cm and represented by a dashed line Water sample images show n when the NMR excitation is optimized for maximum signal near the SC in E and when th e excitation is optimized for maximum signal at the location of the tissue construct (1 cm away from the SC) in F
74 3 4 2 Implantable Coil Coating 3 4 2 1 P olystyrene Every PS layer added an average shift of the resonant frequency of the coil of 2.33 (0.48%) MHz (Table 3 2). The Q did not vary significantly when more layers w ere added. However, both the resonan t frequency and the Q of the coil v ar ied largely when the PS coated coils were put in a phantom and left in there over time (up to 15 hours). The resonan t frequ ency shifted on average by 150 MHz (or 32.4%) when embedded, and by another 100 MHz (22.1%) after 15 hours. The Q dropped 4 -fold after the coil was embedded in the phantom, and could not be measured anymore after 15 hours. Table 3 2. Implantable coil c haracteristics when coated with PS. f: resonan t frequency and # of PS layer: number of PS layers coating solenoid implantable coils. Coil # # of PS layer f (MHz) Solenoid IC (n = 2) 0 489.25 3.89 ( 0.79%) 1 486.50 6.36 ( 1.31%) 3 484.00 2.83 ( 0.58%) 4 479.25 4.60 ( 0.96%) 5 477.25 3.89 ( 0.81%) 6 474.75 1.06 ( 0.22%) 7 475.25 2.47 ( 0.52%) 3 4 2 2 P olytetrafluoroethylene The thickness of each layer of PTFE never exceeded 2 3 m with the 6% PTFE solution when performing the dip coating. It also took a pproximately five hours of continuous labor to add 3 layers onto one coil. In addition, bubbles were present in large number. When the PTFE block -drying was attempted to get a thicker coating in a short amount of time, a skin form ed at
75 the top of the solution. This skin blocked the evaporation of the solvent trapped underneath it and had to be broken apart midway through the drying process to allow the PTFE to continue gelling. The beaker was then left overnight in the hood to al low the solution gel completely. A number of bubbles were present in the 4mm thick layer. However, a coil was placed on top of this first layer along with more PTFE solution B y the end of the second layer procedure the coil had moved from in between t he two PTFE layers to the lower edge of the first layer at the bottom of the beaker. This occurrence ma de the coating incomplete as the coil was never fully covered by PTFE 3 4 2 3 Polymethylmethacrylate and polydimethylsiloxane Coating materials selection: The coating of the implantable coil with PMMA and PDMS were successfully achieved because of their eas y and fast preparation and handling of the se two materials (Figure 3 8). Both coatings could easily be made bubble -free. The PDMS curing in an oven was avoi ded due to the increased risk of air bubble entrapment. A small hollow punch was used to create the construct cavity at the center of the coil (1 -cm diameter) and a larger one was used to give the outer shape to the coated coil including the capacitor (2 -cm diameter) Figure 3 8. Photograph of the successfully coated implantable coils. A) W ith PMMA; and B) with PDMS. (Photo by N.A. Volland).
76 A 1 -mm coating layer was used to coat the coils afterwards to d etermine the changes in the coil characteristic s due to the coating and make the final coating material selection. A resonant frequency shift of 8.13 1.09 ( 13.35%) MHz was detected for the PDMS coated coils compared to the non-coated loop gap resonator (n = 6). This shift was in the same order of magnitude ( 11.3 MHz n = 1) for the PMMA -coated coil tested The quality factor of these coils was only slightly different (214.50 20.95 ( 9.77%) for PDMS -coated coils (n = 6) and 169.7 for PMMA -coated coils (n = 1)) from their non-coated counterpar ts (< 10%) However, t he Q of the inductively-coupled implantable coil system with the PDMS -coated implantable coil was more than twice the Q of the inductively coupled implantable coil system with the PMMA coated implantable coil (104.82 3.22 ( 3.07%) (n = 4) vs. 49.5 (n = 1)) a couple of days only after the implantable coils were coat ed Signs of oxidation were observed at that time on the copper part of the coil (green color) with the PMMA coating Nonetheless, both the PMMA and PDMS -coated implantable coils were tested in the magnet, as t h eir resonan t frequency shift d id not create any problems to tune the inductively coupled coil system ( surface and implantable coils coupled together) at 470.75 MHz The SNR determined from the images of the water sample acquired using the PDMS -coated coils was 29.0% ( 5.63%) better than the SNR determined with the PMMA -coated coil. These results made PDMS the coating materials of choice for the implantable coil in all the future studies of this project Further considerations of these results are detailed in Section 3.5 Coating thickness determination: The effects of coating on the coil were further characterized next and the optimal coating thickness was determined. Figure 3 9 illustrates changes in coil characteristics as the thickness of the PDMS around the coil is varied (n = 2 for each coating thickness). The coating caused the coil resonan t frequency to shift down (Fig ure 3 -
77 9A black triangles). However, the Q of a PDMS -coated coil is similar to that of a non coated coil, (Fig ure 3 9B black triangles). When a PDMS coated coil is loaded (e.g., in a gel phantom), an additional negative frequency shift occurs (Fig ure 3 9A grey diamonds) and the Q decreases (Fig ure 3 9B grey diamonds). The frequency shift is caused by the addition of capacitance to the coil circuit Even though Q was determined at a different frequency, t he change s in Q can be attributed to the addition of resistance fro m the coating and the sample. If the coating thickness altered the frequency slightly, with a maximum shift approaching 10 MHz (Fig ure 3 9A black squares), upon immersion, the frequency shift decreased as the coating thickness increased, with only a 3% drop in the coil containing an 8 -mm PDMS coating (Figure 3 9A open diamonds). Moreover, having a thicker coat reduced the capacitive coupling of the coil to the sample and reduced the effect of sample loading on the coil Q. Further considerations of the se results for t he final selection of the coating thickness and the practical limitation of construct fabrication are detailed in Section 3.5 However, a 1 -mm coating thickness was chosen as the optimal coating thickness for the purpose of the work presented here. Coated implantable coil characterization: With the thickness of coating set, PDMS coated implantable coils could be generated and fully characterized prior to in vitro and in vivo testing. The large hollow punch used to define the outer edge of the coating coil had a 2 -cm diameter. The PDMS left in the center of the coated coil was removed with a 1 -cm diamater hollow punch. As previously reported in S ection on Coating materials selection the 1 -mm PDMS layer created a 8.67 1.12 ( 12.97%) MH z (n = 4) frequency shift in the coil resonant frequency. In addition, loading further changes the coil characteristics.
78 Figure 3 9. Effects of PDMS coating thickness on the implantable coil characteristics A) F requency shift of an unloaded IC ( ) and a loaded IC () versus coating thickness ; and B) % change of an unloaded IC Q ( vertical scale on the right) and a loaded IC Q (, vertical scale on the left ) versus coating thickness The right -side vertical scale compares the ratio of unloaded coated coil Q (QUC) to unloaded, uncoated coil Q (QU U) and the left -side vertical scale compares the ratio of coated, loaded coil quality factor (QL C) to coated, unloaded coil quality factor (QUC). A photograph of the different coated IC tested is shown b etween graphs A and B.
79 When the PDMS -coated implantable coils were embedded in the mouse abdomen like phantom, an additional resonan t frequency shift from the PDMS coating of -25.70 1.79 ( 6.95%) MHz (n = 4) was observed. These two successive shifts ma de it impossible to tune the coupled coil system to the desired frequency of 470.75 MHz and w ere taken into account to build the implantable loop-gap resonator coils which were used for in vitro and in vivo studies. Hence, the implantable coils to be used loaded w ere constructe d with a resonant frequency approximately 34 MHz higher than their predecessors (530.23 3.71 ( 0.70%) MHz, n = 4 ). This made the inductively -coupled coil system s resonate at the desired frequency of 470.75 MHz after coating and lo ading into the gel phantom or in vivo and ready to be tested in the magnet. The i mplant able coils resonated at 497.64 3.58 ( 0.72%) MHz when embedded 0.42 0.02 ( 5.31%) cm deep in the phantoms and at 492.93 3.44 ( 0.70%) MHz when embedded 0.74 0. 06 (8.46%) cm deep in the phantoms (n = 4). The quality factor of the loaded implantable coils was affected by the loading and decreased by a factor of 13 (18.51 3.99 ( 21.57%)) when embedded 0.42 cm in the phantom and by a factor of 16 (15.41 0.79 ( 5.12%)) when embedded 0.70 cm in the phantom implanted in the peritoneal cavity of a mouse The Q of the inductively-coupled implantable coil system was dropped down to 9.39 0.32 ( 3.45%) MHz when the PDMS -coated implantable coil was embedded 0.70 cm deep in the gel phantom and to 10.96 0.33 ( 3.02%) MHz when the PDMS -coated implantable coil was embedded 0.42 cm deep in the gel phantom. 3 4 3 Coil-Construct Assembly The three constru cts were successfully generated and placed in the middle of a PDMS coated implantable coil as presented on Figure 3 10. These implantable coil -construct assemblies were all k ept intact when soaked into HBSS for over 2 weeks.
80 Figure 3 10. Photograph of the i mpl antable coil -construct assemblies considered in this study A ) A garose gel construct; B) alginate foam construct; and C ) alginate beads hold in the middle of the coated coil by two meshes. Every construct has been dy ed with trypan blue to help their localization. (Photo by N.A. Volland). However, the agarose constru ct did not hold very firmly at the center of the coated coil and was also brittle. This could create a problem when the construct was implanted in vivo as it may easily separate from the implantable coil simply with the regular movement of the host and was not conside re d any further for the purpose of this study The alginate foam construct was later found to have pores too large to properly host cells within the construct as it is shown in Chapter 6 and was not implemented further either in this study. C onsequently, the alginate beads hold together by a mesh glued on a coated coil was the construct selected for th e purpose of th is study. Cells could be kept alive within the construct (Chapter 6) in the middle of the coil to allow their direct and non -inv asive monitoring with NMR methods. 3 4 4 P hantom S tudies Figure 11 illustrates differences in SNR gain for different configurations and types of coils studied under unloaded conditions. The RF coils compared in this study include: 1) an SC at both 4.7 T and 11. 1 T (n = 1 for each field); 2) the two resonance modes of an overcoupled system with an SC and a solenoid IC at 11.1 T (n = 5 for the (+) mode and n = 4 for the ( ) mode); and 3) the two resonance modes of an overcoupled SC and a loop-gap IC at 11.1 T (n = 7
81 for the (+) mode and n = 4 for the ( ) mode). The implantable loop gap resonator in the (+) mode was coated with a 1 -mm PDMS layer and tested when inductively coupled with a SC (n = 4). Figure 3 11. SNR comparison between different RF coils tested under unloaded conditions using a water sample The coils tested were: an SC at 4.7 T (SC 4.7T) and at 11.1 T (SC), coupled -coil systems with a solenoid IC or a loop -gap IC in the ( -) mode (Sol & LG ) and in the (+) mode (Sol+ & LG+), or a PDMS coated lo op -gap in the (+) mode (LG+ C) placed 1 cm away from the SC PDMSsup indicates that a PDMS suppression schema was tested f or LG+ C denotes the statistical difference between the coil SNR at 11.1T and the SC SNR at 4.7 T ; # the difference between the c oupled coil system SNR and the SC SNR at 11.1T; the difference between the coupled coil system SNR in the (+) and ( ) mode ; and + between the SNR of LG and Sol -. The SNR of SC (SC) was better at the higher magnetic field strength of 11.1 T (Figure 3 11) approximately 4 times higher than at 4.7 T (SC 4.7T) [4. 07 0.7 0 ( 17 .2 %), p << 0.001]. At 11.1 T all coupled-coil systems performed better than the SC. The SNR of the coupled-coil system in the ( -) mode was approximately 3 times greater than the SC (SC 11T) [12.1 2.27 ( 18.8%), p << 0.001] when the IC was a solenoid (Sol ) and 3.4 times greater [14.0 2.40 ( 17.1%), p << 0.001] when the IC was a loop-gap (LG ). The (+) mode of these coupled -coil systems surpassed the performances of the ( ) m ode. The coupled-coil system with a solenoid IC
82 (Sol+) was approximately 5.3 times greater than the SC (SC 11T) [21.4 3.06 ( 14.5%), p << 0.001] and the system with the loop gap resonator IC (LG+) nearly 5.2 times greater [21.1 3.27 ( 15.5%), p << 0.001]. Therefore, the (+) mode coupled -coil systems outperformed the ( ) mode coupled -coil systems when either a solenoid (p << 0.001) or a loop gap (p << 0.001) were used as IC. The coupled coil system with a loop-gap resonator IC performed better than the system with a solenoid IC in the ( ) mode (p << 0.001), but was not statistically better in the (+) mode (p>0.05). The coupled-coil system in the (+) mode with a PDMS coated loop gap resonator IC also showed an SNR improvement of approximately 5 times that of the SC (LG+ C) [20.3 3.20 ( 15. 8 %), p << 0.001]. Furthermore, PDMS suppression used with the implantable coated coil, did not significantly affect the inductively -coupled coil system SNR (LG+ C PDMSsup) [ 18. 7 2.9 4 ( 15. 7 %), p << 0.001] Fig ure 3 7 illustrates SNR differences between the (+) and ( ) modes The ratio of B1 between the (+) and ( ) modes for similar coils approaches 1.5 as shown by comparing the ratio of SNR for the inductively -coupled coil system in the (+) to ( ) mode first, with the loop -gap IC [1.51 0.35 ( 23.1%)] and second, with the solenoid IC [1.76 0.42 ( 26. 6%)] The magnetic field distribution of a SC is also represent ed for comparison (Figure 3 7D & 3 7E ). The nonuniformity of the image in Figure 3 7F is due to the overtipping of spins necessary to obtain the proper excitation at the implanted construct location. These simulations indicate that the B1 field magnitude is highest at the IC location in the (+) configuration (Figure 3 7I & 3 7J). The ratio betw een the simulated B1 field magnitude at the construct location of a SC and an inductively coupled coil system in the (+ ) mode is 4.5 (Figure 3 7D & 3 7I ). This ratio agrees reasonably well with the SNR ratio of ~ 5.2 between the measured SNR at the constr uct location for a SC and a coupled -coil system in the (+ ) mode for the water and gel phantoms (Figure 3 11 & 3 1 2 ).
83 Figure 3 12. SNR comparison between different RF coils tested under loaded conditions in a gel phantom. An SC at 4.7 T (SC 4.7T) and a t 11.1 T (SC), and a coupled-coil system with a PDMS -coated loop -gap IC in the (+) mode placed 1 cm away from the surface (LG+ C) were tested. PDMSsup indicates that a PDMS suppression schema was tested the PDMS -coated IC. denotes the statistical diffe rence between the SNR of the RF coils at 11.1T and the SC at 4.7 T); and # the difference between the coupledcoil system SNR and the SC SNR at 11.1T. Because the coupled coil system in the (+) mode gives the highest SNR under unloaded conditions (Figure 311) when either the solenoid or loopgap IC was used, the effect of loading on the measured SNR for a coupled -coil system were investigated in the (+) mode only. The loop -gap IC design was selected for ease of construction. Figure 31 2 depicts the SNR g ain obtained at the construct location (approximately 1 cm away from the SC) when coupled -coil systems comprising PDMS -coated loop gap IC (implantable -coil -construct assembly) and SC were tested under loaded conditions (n = 4). With an SC, the SNR increas e factor at 11.1 T compared to 4.7 T is significant [2.38 0.27 ( 11.4%), p<<0.001]. At 11.1 T the coupledcoil system shows an SNR increase of more than 5.25 times that of the SC [ 12.5 2.6 1 ( 20.9%), p << 0.001] when the implantable coil construct a ssembly was placed approximately 1 cm away
84 from the SC [0.74 0.06 ( 8.46%) cm]. The PDMS suppression did not significantly alter the results [ 12. 5 2. 85 ( 22. 9%), p<<0.001]. Figure 3 13 displays the SNR gain obtained when the distance between the i mplantable coil -construct assembly and the SC approximated 0.5 cm (0.42 0. 05 ( 10.6%) cm in a gel phantom and 0.62 cm 0.08 ( 18.8%) in vivo ). Th e sensitivity improvement reduces to a factor of ~1.7 with the reduced distance between the two coils of t he system [4.03 0.90 ( 22.3%) p << 0.001 in a gel phantom and 4.34 1.51 ( 34. 8 %), p << 0.001 in vivo ]. The PDMS suppression did not significantly alter the results [4.02 0.88 ( 22. 4%), p<<0.001 in a gel phantom and 4.86 1 .6 0 ( 32.9 %), p << 0 .001 in vivo ] at this separating distance either. Figure 3 13. SNR comparison between different RF coils tested under loaded conditions using a gel phantom or in vivo conditions at 11.1T Were tested: an SC (SC) and coupled coil systems with a PDMS -co ated loop -gap IC in the (+) mode placed 0.5 cm away from the surface in the phantom (LG+ C) or implanted in vivo (LG+ C in vivo ). PDMSsup indicates that a PDMS suppression schema was tested. # denotes the statistical difference between the SNR of the ind uctively -coupled coil systems and the SC.
85 3 4 5 In vivo S tudies in M ice Bioartificial pancreatic construct s containing PDMS -coated loopgap coil s w ere successfully implanted into m ice peritoneal cavit ies (n = 3). The implant able coil s w ere found to resonate at 5 01.59 2.63 ( 0.53%) MHz when in the peritoneal cavity of mice Q was found to have decreased by a factor 11.5 (20.08 3.73 ( 18.58%)) when implanted in the peritoneal cavity of mice compared to the coil Q before implantation The inductively -coupled implantable coil system Q was now 35.25 10.25 ( 29.08%) MHz when the implantable coil was in the mouse peritoneal cavity Constructs without the implantable coil but with the same characteristics than the constructs including a coil have also been impl anted successfully (n = 3). The surface coil Q was 61.78 5.78 ( 9.35%) MHz when the implantable coil was in the mouse peritoneal cavity. In vivo images (Fig ure 3 1 4 ) were acquired within 1 hour following implantation using conventional respiratory gati ng and a n effective TR of 2500ms. These images provided a clear visualization of the construct. If i ncreasing the number of averages and the number of averaged experiments reduced the animal breathing artifact on the images these two improvements increa sed the acquisition time and were consequently not used in this study However, better respiratory gating without an significant increase in acquisition time could be investigated in the future: self -gating (155157) or an image reconstruction using a motion artifact correction post processing imaging gating (158) The inductively -coupled coil system shows better localization of the construct than the SC (n = 3), because it is mo re spatially selective. The average distance from the implanted construct to the mouse skin was 0.42 cm [ 0.08 (18.8%)] making separation between the IC and the SC about 0.6 cm, when accounting for the mouse holder. The inductively -coupled coil system
86 shows an SNR increase of approximately 2 times over an SC at 11.1 T [4.34 1.51 ( 34.8 %), p << 0.001, without PDMS suppression; 4.86 1.60 ( 32.9%), p << 0.001, with PDMS suppression] (Figure 3 13). With this sensitivity improvement, in vivo images o f the construct allowed distinction of individual alginate beads, 400to 700m -diameter, entrapped within its cavity (Fig ure 3 14B ) with an in -plane resolution of 78.125 m, demonstrating the potential of this inductively -coupled implantable coil system t o directly and noninvasively monitor a bioartificial pancreas in vivo Figure 3 14. Cross -sectional images of two bioartificial constructs implanted in vivo A) Plain construct imaged using an SC; and B) an implantable coil -construct assembly (with t he same characteristic as the plain construct) imaged using the coupled -coil system. The white circle delineates the construct outer edge, visible in A, but not in B. An SE pulse sequence was used to acquire these images: TR = 2 500 ms, TE = 10 ms, 1 -mm s lice thickness, 1 average FOV 2 x 2 cm2 and matrix size 256 x 256. Respiratory gating was also applied. 3 5 Discussion NMR is a powerful technique for acquiring images and spectroscopic data, but its inherent insensitivity limits application of this techniqu e when studying implanted bioartificial constructs. To overcome this limitation, the use of an inductively-coupled, implanted coil system for NMR
87 measurements was explored This system has substantive advantages over an SC, and a working in vivo implanta ble coil system that can be implanted with the bioartificial c onstruct was developed. For the study of tissue -engineered constructs, this inductively-coupled, implantable coil system has the following advantages: 1) measured NMR signal is more homogeneous throughout the construct; 2) the coil system is more spatially selective without the use of selective pulse sequences because the IC surrounds the construct; and 3) if implantation depth is variable, the inductively-coupled coil system allows tighter cont rol over RF coil tuning and matching, and the B1 field distribution is more uniform. Additionally, when applied in vivo with respiratory gating, this system is less sensitive to breathing artifacts: the implantable coil system resonant frequency shift is small with motion, whereas with an SC the loading and sensitivity at the construct location changes more with motion. The available SNR in the NMR measurement is a primary concern when obtaining useful information from an implanted construct. Increasing the static magnetic field strength improves sensitivity. Under unloaded conditions, the SNR is proportional to (B1/B2)7/4, where B1 and B2 are the magnetic field strengths being compared (B1 > B2) (139) and primarily due to the higher sensitivity obtained at high er field (97) This predicted enhancement in SNR of 4.5 times between 4.7 T and 11.1 T is similar to the measured SNR improvement of 4.07 0.18 ( 4.34%) when SC were tested. Under loaded conditions (like tissue over a surface coil and surrounding an implanted coil), the loading reduces the SNR gain to a linear increase with magnetic field strength (7). For field strengths considered here, the predicted SNR improvement is 2.36 -fold agreeing well with the measured valu e of 2.38 0.27 ( 11.4%). The inductively -coupled, implanted coil system also demonstrates an SNR improvement over a single SC at 11.1 T from a minimum of about 1.7 to over 5, depending on the implantable
88 coil location. The inductively -coupled coil syst ems in (+) mode with both the solenoid and loop gap resonator IC yield better SNR as shown on Figure 3 11A These systems were chosen for NMR measurements. Although the inductively-coupled coil systems in the (+) mode with the solenoid IC gave slightly b etter SNR than the inductively coupled system with the loop-gap IC, the difference is not statistically significant. The solenoid IC require small series capacitances for both (+) and ( ) modes, due to the high inductance of the solenoid design, making the system harder to build and compensate for coating and loading effects. Since the loop-gap coil has lower inductance, which requires larger series capacitance, it is easier to build and coat than the solenoid. Therefore, these important considerations ( lower inductance and ease of construction) make the loop-gap resonator IC the best overall choice to build bioartificial constructs. With regards to coating, the frequency shift and the decrease in quality factor of the implantable over time observed wit h PS coating was explained by the PS layer tendency to absorb water over time (136) and thus modifying the coating properties. This evolutionary behavior was not acceptabl e for the purpose of this study PS was consequently disregarded. PTFE was set aside as it was hard to handle even though it has good dielectric properties. Furthermore, PTFE is a thermoplastic polymer, not a thermosetting polymer (crosslinked polymer f ixed once and for all). This means that if it is reheated above its glass temperature, it will melt again. This was the explanation for the coil falling to the bottom of the beaker repeatedly. The bottom layer of PTFE melted making the coil fall to the bottom of the beaker. The two PFTE layers then became one. Injection and/or compression molding, and chemical vapor deposition (CVD) procedures could have be en considered. However, more attention will have to be given to these techniques as they have hi gh temperature (250400 C) and/or high pressure (4 7 MPa or 5001,000 psi) requirements PMMA coating was a bandoned since images
89 acquired using the implantable coils with this coating displayed a lower SNR compared to the ones acquired using implantable coil with PDMS coating The coil oxidation that appeared within days of the coil coating may have been the reason for the SNR deterioration. I t was also slightly harder to handle than PDMS. For this study, o nly PDMS met all the criteria to easily and ef ficiently coat the implantable coils and was used for the in vitro and in vitro studies. T he PDMS coil coating induces mainly capacitance losses shift ing the resonan t frequency while coil loading mainly introduce s resistive losses, decreas ing the Q of the system (see Figure 3 9). Taking into account these consequences, these results suggest that an 8 -mm coating would minimize changes to both the Q and frequency shift of the IC. However, the resultant implantable coil -construct assembly is unrealistic all y large for monitoring the construct in viv o The size of the mouse peritoneal cavity limits the coating thickness to no more than 1 mm. Furthermore, the insertion of the coil around the construct requires the coil to be close to the construct to maximiz e sensitivity (filling factor). Therefore, a PDMS coating thickness of 1 -mm is the best compromise to maximize coil function while allowing sufficient space for a viable construct (i.e., sufficient biological component of a bioartificial organ). The indu ctively -coupled coil system with the coated loop-gap IC was successfully developed, tested in a mouse abdomen like gel phantom, and impl a nted in a mouse and observed in the (+) mode Although the SNR decreases under loaded conditions compared to the unloaded conditions, there is sufficient SNR to obtain better images than with a SC. The coated loop gap system was successfully developed and tested in a phantom (loaded conditions similar to in vivo conditions), as well as implanted in a mouse and observed i n the (+) mode. Although the SNR decreases under loaded conditions (mouse abdomen-like phantom or
90 actual mouse) compared to the unloaded conditions, there is sufficient SNR to obtain better images than with a surface coil. For in vivo imaging, only respiratory gating and increased TR close to the breathing rate of the animal were kept to image the bioartificial pancreas as the best compromise between acquisition time and SNR. The 400to 700m -diameter beads within the construct were visualized in vivo (Fig ure 3 1 4 ) with an in -plane resolution of 73.13 m, demonstrating the potential of this inductively -coupled implanted coil system to directly and non -invasively monitor a bioartificial pancreas in vivo With fixed coil sizes, the advantage of an induct ively -coupled implanted coil over an SC will increase as the depth of the implanted coil increases, such as in larger animal models (e.g. humans), since the SC sensitivity drops rapidly with distance (Fig ure 3 7D & I ). If a larger SC can obtain a greater depth of sensitivity for NMR measurements, the noise sensitivity would increase with the SC sensitive volume. Therefore, the inductively-coupled implanted coil presents an optimum configuration for NMR detection from implanted tissue constructs, though no t restricted to use with implanted tissue -engineered constructs. 3 6 Summary Implantable inductively -coupled loopgap coils assembled within bioartificial constructs have been constructed, implanted into mice, and their performance analyzed. NMR images from t hese assemblies were easily obtained at 11.1 T, demonstrating the capability of this inductivelycoupled implanted coil system. The sensitivity improvement of this implantable coil system over SC at 11.1 T is approximately 2 -fold. This gain was multiplie d by the improvement of 2.4 -fold obtained from the magnetic field strength increase from 4.7T to 11.1T. The sensitivity improvement available with the use of inductively-coupled, implantable coil system allows significant gains in information obtained from an implanted construct, providing images with
91 higher contrast to noise, spectroscopy with greater SNR (e.g., 1H observation of choline for cell containing constructs), and allowing detection of other key, but less -sensitive nuclei (e.g., 31P, 19F) critic al in the efforts to quantify implanted bioartificial organ function. It is important to emphasize that the approach outlined here can be applied to other situations, and is not restricted to use with implanted tissue -engineered constructs, although its i nherent strength make it a desirable choice when obtaining information from implantable constructs. With the technology of implanted coil established, this research focused then on perfecting NMR acquisition parameters (e.g. eliminating artifacts, due to h ardware and physiology issues); developing a receive -only inductively -coupled implantable coil system that will further enhance the ability to perform 1H NMR localized spectroscopy, and studying cellularized construct s through NMR imaging and spectroscopy
92 CHAPTER 4 DEVELOPMENT OF A RECEIVE ONLY INDUCTIVELY COUPLED RF COIL SYST EM TO ENHANCE 1H NMR LOCALIZED SPECTROSCO PY TO MONITOR AN IMPLANTABLE BIOARTIF ICIAL CONSTRUCT 4 1 Introduction The monitoring of bioartificial pancreas (BAP) viability is of primary importance a s it directly reflects its well being and functioning. Choline is a vital amine molecule present in every living cell, whose NMR signal can be detected when performing 1H MR spectroscopy (MRS) and has been linked to the number of viable cells within the a rea of detection. To allow optimal MRS, a uniform NMR excitation and a highly sensitive NMR response signal detection is recommended. In order to achieve this optimum, a receive -only inductively-coupled, implanted coil system was developed in this study. Its use was demonstrated at high field (11.1 T) in combination with a transmit -only birdcage volume coil. Its specific requirements for monitoring a bioartificial construct in vitro were also investigated. The results establish that even though no overa ll gain in signal -to -noise was obtained with this receive -only inductively coupled implanted coil system over that which can be obtained with a transmit -receive inductively-coupled implanted coil system, this coil system provides a greater signal uniformit y and a means to analyze the function of implanted bioartificial organs quantitatively using choline NMR detection. 4 2 Background The monitoring of the viability of cells within a tissue -engineered construct, such as a bioartificial pancreas (BAP), is of prim ary importance as it directly reflects the well being and functioning of the construct. NMR has been shown to be well -suited for monitoring a bioartificial pancreatic construct noninvasively and temporally, in vitro and in vivo because of its penetratio n depth, resolution, contrast, and ability to probe both metabolism and anatomy (2 -
93 4) While NMR imaging mainly provides anatomical information Stabler et al. developed an NMR -based method to noninvasively assess the viability of cells within a bioartificial pancreatic construct using 1H MR spectroscopy (MRS) (5,6) and detecting the NMR choline signal. Choline is a vital amine molecule, which is present in every living cell. It participates in cell -signaling ser ves as an acetylcholine precursor, and adds structural integrity to cell membranes. The choline NMR signal, also referred to as total choline (TCho) signal, includes the signal from the following aqueous compounds found intracellular ly : phosphocholine (PC ho, a precursor of membrane phospholipids), glycerophosphocholine (GPCho, a product of the degradation of membrane phospholipids) and free choline (Cho) (159) Their contribution to the total choline signal has been determined at 64% for PCho, 33% for GPCho, and 3% for Cho (88) The signal can be detected when performing 1H MR spectroscopy (MRS). Even though the c ho line molecule ha s several resonance peak s on a 1H spectrum as shown on Figure 23B its strongest one can be found at 3.2 ppm due to the molecule s three equivalent methyl groups. Variations in th e signal intensity of the T Cho signal at 3.2 ppm have been show n to represent the increase of membrane phospholipid biosynthesis and serve as an active marker for cellular proliferation in tumors (86,87) By localizing the spectroscopy to wit hin the BAP, TCho intensity changes have been shown to be linked to oxygen concentration (88) and proportional to cell viability variation (5,6) These later studies clearly demonstrate that acquisition of metabolic information from cells imbedded with in tissue -engineered constructs implanted intraperitoneally in mice abdomen is possible by MRS Unfortunately, the sensitivity of the method used was limited by the strength of the magnetic field and the radiofrequency (RF) coils used to excite and detect the signal (a 4.7 T NMR instrument and a surface coil ). T he threshold of NMR signal
94 detection due to t hese experimental conditions were approximately one million cells per construct under in vitro ( ideal ) conditions and two million cells under in vivo conditions. Consequently, the constructs were generated with a large number of cells and could st ill only be monitored for a maximum of two weeks due to the lack of sensitivity of the monitoring method. Recently, the use of higher magnetic field strength along with inductively-coupled implantable coil systems was shown to greatly improve the sensitivi ty of this method ( Chapter 3 and (160)). This sensitivity improvement would allow the detection and study of lower density of cells within constructs in vitro and in vivo over a longer period of time. Furthermore, the selectivity of inductively -coupled implantable coil system demonstrated to be superior to the surface coil alone ( Chapter 3). This gi ves an advantage to inductively -coupled implantable coil system when performing localized spectroscopy with a high sensitivity as it decreases the signal contamination from outside of the constr uct. H owever, this system does not produce a homogeneous RF magnetic field over a large enough volume to use volume selective pulse sequences to their full extent (81) In order to optimize the power of selective pulse sequences, a uniform excitation magnetic field pattern within the sample is preferable. L arge surface coils or volume coils can achieve such an excitation pattern, though decreasing the signal detection sensitivity. Combining a volume coil and an inductively -coupled implantable coil system closer to the area of interest allows production of a uniform excitation and d etect ion of the NMR signal with a higher sensitivity. However, the two systems have to be decoupled from each other to prevent the ir interact ion and allow the proper functioning of this two RF -coil setup The volume coil was actively decoupled by a n external bias current source (104,106) since the NMR signal from the sample w ould not be high enough to generate the power required to make the transmit only coil decoupling circuit function properly. U nlike the volume coil, the inductively -coupled
95 coil system ( implantable and surface coils included ) was passively decoupled (105) The use of a passive decoupling avoid ed the requirement for external diode bias connection to the implantable coil running through the skin. This report investiga tes the use of receive -only inductively-coupled RF coil systems at a high field (11.1 T) to monitor implanted tissue -engineered constructs with NMR and perform 1H localized spectroscopy to detect choline signal. It also compares NMR sensitivities for moni toring an implanted bioartificial pancreatic construct using either an inductively-coupled implanted coil s ystem for both excitation and detection or the combination of a birdcage volume coil for excitation and a receive -only inductively -coupled implanted coil s ystem for detection T his report also discusses technical aspects important toward generating receiveonly inductively-coupled coil systems for 1H detection using passive decoupling, and issues to consider when applying these coils in conjunction wi th implanted bioartificial constructs for localized spectroscopy. 4 3 Materials and M ethods 4 3 1 RF Coil S ystem D evelopment All the inductively-coupled implantable coil systems used in this study were constructed tested and compared for NMR imaging and spectroscopy at 11.1 T Because of the results obtained in Chapter 3, t h ese coupled -coil system s consist ed of an internal (implantable) loop -gap resonator coil inductively coupled to a surface coil The y were also used when overcoupled in their co rotating mode (s ee S ection 3.3. 3 for more details on inductive coupling and overcoupling). T he two different inductively -coupled systems developed were: 1) one system function ing a s both transmitter and receiver; and 2) one system function ing as receiver only for 1H NMR (470.75 MHz at 11.1 T) Th is second system was coupled with a transmit -only home built bird -cage volume coil providing the NMR excitation. The volume coil was actively
96 decoupled from the inductively -coupled implantable coil system when this latter system was detecting the sample signal whereas the inductively-coupled coil system was passively decoupled from the volume coil when the volume coil was exciting the sample The first inductively-coupled system was the system used in the previous study present ed in Chapter 3 as well as t he i mplantable coil construct assembly (160) To render the second inductively -coupled implantable coil system receiveonly, the addition of a passive decoupling circuit was required on both the surface and implantable coils (104,105) This circuit was composed of anti -parallel diodes (D1s and D2s for the surface coil and D1i and D2i for t he implantable coil on Figure 4 1) and a small solenoid inductor (LD s for the surface coil and LDi for the implantable coil on Figure 4 1) in parallel to one of the tuning capacitors on the resonant circuit ( American T echnical Ceramics, Hartford, CT; CT s for the surface coil and CTi for the implantable coil on Figure 4 1). Figure 4 1. Circuit diagram for a receive only inductively -coupled implantable coil system for 1H detection A ) I mplantable coil; and B) surface coil. Inductors (L), capacitors (C), and diodes (D) are shown with subscripts denoting implantable (i) or surface (s ) coil Capacitors are also designated as tuning (CT) or matching (CM) as well as inductor as main circuit inductor (L) and decoupling inductor (LD). M indicates the mutual i nductance or inductive coupling and V variable A cable trap equivalent, including an inductor and a capacitor, is also shown in the diagram.
97 The diode had an active (ON) and an inactive (OFF) state that could be modeled with equivalent circuit s as follow s : 1) when the diode was in the OFF position allowing the coil to function as a detector the diode was equivalent to a capacitor in parallel with a high value resistance (leakage resistance), no current was flowing in the decoupling circuit the receive -o nly coil functioned as if the decoupling circuit was not there; and 2) when the diode was turned ON, the receiving coil was prevented from functioning. The diode was then equivalent to a small series resistance and effectively switched the decoupling circ uit inductor in parallel with the capacitor, creating a high impedance in the coil circuit. This impeded the current flow in the coil circuit making the receive-only system invisible to the transmit -only coil. In the magnet, this circuit wa s activated by the RF excitation signal from the transmit -only volume coil and d id not require the use of a pulsed DC control signal source (as does an active decoupling circuit). To achieve this decoupling at the coil resonant frequency, the impedance of the decoupling circuit has to be equal to the impedance of the capacitor in parallel. This means that the decoupling inductors impedance equals the tuning capacitor i mpedance (105) when the diode equivalent series resistance can be considered small enough The decoupling circuit solenoid was made of 25 AW G (American Wire Gauge) copper wires hand wound around a 16 AWG copper wire. Its value was verified using the RF impedance test adapter HP 1620A spring clip fixture of the network analyzer ( 8752C Hewlett Packard Santa Rosa, CA) before being placed perpen dicular to the main coil loop to limit the interaction between each other Different inductively -coupled implantable coil system layouts were built and tested to optimize the design. These layouts differed from each other mainly by the type of diodes us ed and their characteristics (mainly their parallel capacitance, equivalent series and parallel resistance and transition time values ) and the number of capacitors the coils comprised. The
98 diodes used were: 1) Schottky fast recovery small signal anti -par allel diodes ( M/A COM MA45471, Tyco Electronics, Berwyn, PA); and 2 ) PIN diode (DH8005240N, Chelton Telecom & Microwave, Grigny, France). The number of tuning capacitor was varied from one to four in the implantable coil circuit, and from two to four in the surface coil circuit. 4 3 2 Coating Techni ques The material used to coat the receive -only implantable coil was the same as the one used for the transmit -receive implantable coil: polydimethylsiloxane (PDMS, Medical Grade Silicone Elastomer, Factor II, Lakesi de, AZ) The technique applied to coat the coil has been described in details in S ection 3. 3.2.4 (160) However, a daptations had to be made to th is coating procedure to accommodate th e receive -only coil extra components. Instead of generating a 1 mm PDMS layer first, a 0.5 -mm one was gene rated to compensate for the slightly wider than 2 -mm diode. A 4 -mm layer was then added on top to cast the coils and make sure the coating was properly sealed all around the coils. Over 24 hours the PDMS would flow in the gap between the coil and the firs t PDMS layer. The outer edge of the coated coil was shaped using a 2.6-cm -diameter hollow punch instead of the 2-cm -diameter one used in the study presented in Chapter 3, since the decoupling circuit made the coil outer edge larger. Furthermore, the coat ing of the handwound decoupling-circuit solenoids was tested prior to the coating of the entire coil as they could be easily deformed under pressure resulting in drastic cha nges in their characteristics (inductance in this case). 4 3 3 RF Coil Testing In order to have the receive-only inductively-coupled coil system resonating at 470 .75 MHz, the implantable and surface coil were individually tuned and matched at the same higher frequency (490.1 MHz) as the transmit receive system after the decoupling circuit was added in
99 their circuit. T he capacitors CM 1V, CM 2V, and CT sV (Fig ure 4 1 ) of the surface coil were adjusted to achieve the match of the whole system to 50 at 470.75 MHz. T he receive-only coils were constructed and optimized on the bench with a network analyzer (Hewlett Packard 8752C, Santa Rosa, CA) before being test ed in the magnet. The quality factor (Q) of each coil taken individually and of the whole system was recorded after construction with and without load and every time the system was tested in the magnet as described in details in S ection 3.3.3. The effects of the decoupling circuit were also tested here by determining changes in the resonant frequency and Q for all the coils and systems (139) after its insertion in the coil main circuit The diff erence in power transmitted to an external field probe located 1 cm away from all the coils when their decoupling circuit was ON and OFF was also determined. A 20 dB difference was at least required for the proper functioning of the decoupling circuit. Th is means that over 90% of the signal from the coil was attenuated when the decoupling circuit was ON compared to when it was OFF. The effects of the 1 -mm coating s on the coils were also tested the same way both in air and in a gel phantom The results ob tained in Chapter 3 (S ection 3.4) for the transmit -receive coupled coil system were used as a reference here. 4 3 4 Samples Receive -only coil systems ( both coated and uncoated) were first tested using a distilled water sample placed in a 10 mm -diameter thin -wall glass tube (Wilmad Labglass, Buena, NJ) on the bench and in the magnet to assess their sensitivity This water phantom was considered to represent the unloaded condition because of its small size compared to the SC and the nonconductive nature of the sa mple. The implantable coils (IC) were placed around this sample no more than a 1 cm away from the surface coil. To test their performances under loaded conditions
100 (similar to in vivo conditions) next the implantable coated coil s were placed in a phantom comprised of a viscous 6.7% (w/w) polysaccharide gel (TX 151; Oil Center Research, International LLC, Lafayette, LA), with the average dielectric characteristics of a mouse abdomen (146148,161) as described in details in S ection 3.3.4. Inductive coupling was accomplished by placing the phantom (with implantable coil) on top of the surface coil. Different types of phantoms were also created for localized spectroscopy testing. First, solutions of different concentration of choline were prepared (100 mM, 10 mM, and 1mM), placed in 10 mm-diameter thin -wall glass tube s (Wilmad Labglass, Buena, NJ) and used to perform 1H spectroscopy and localized spectroscopy. Then, a 1 mM solution of choline was pl aced in a sealed 5 -mm thin -wall glass tube at the center of a 10 mm -diameter thin -wall glass tube filled with glycerin to assess the performance of the localized spectroscopy method when using the two inductively-coupled implantable coil systems under inve stigation in this study under unloaded condition. Then, a 1 mM solution of choline was also placed in the middle of an altered coil construct assembly to test the method using the same system s under loaded conditions (see S ection 4.3.5). In vitro (live cells in a construct) localized spectroscopy was attempted last. 4 3 5 Cell Culture tet cells (48) were obtained from Shimon Efrat s lab oratory (Albert Einstein College of Medicine, Bronx, NY ). They were cultured in T -flasks in monolayers and fed every 2 -to 3 days with high glucose Dulbeccos Modified Ea gles Medium (DMEM, Invtrogen Corp., Carlsbad, CA ), supplemented with 10% fetal bovine serum (FBS, Lonza Williamsport, PA) 1% penicillin -streptomycin (Cellgro, Manassas, VA ) and 1% L glutamine (Cellgro, Manassas, VA) until they reached confluence. The cells were then collected from the flasks for construct fabrication using 0. 01% Tryps in (Sigma, St. Louis, MO)
101 4 3 6 Construct Development A 16-mm diameter biocompatible polyetheretherketone (PEEK) 300-m mesh (Small Parts, Inc., Miami Lakes, FL) was added on t he top and bottom of the ring formed by the PDMS coated coil to create the construct cavity (Figure 4 2) in both implanted coil types (receive -only and transmit receive ). For in vitro studies only, 2mm diameter stainless steel spheres were attached with t he mesh to one side of the PDMS ring to create a stand and help with the homogeneous diffusion of nutrients and oxygen through the construct. Three spheres were necessary for the implantable coil -construct assembly including the transmit receive implantab le coil. Four spheres were added to the assembly including the receive -only implantable coil to increase its stability as it was larger. Figure 4 2. Implantable coil bioartificial pancreas macroconstruct assembly schematic and dimensions for the tran smit receive implantable coil A) Transverse view and B) coronal view of the assembly. Some of these coil -construct assemblies were altered to receive choline solution instead of cell -containing beads by replacing the 2 meshes of the construct by 2 pieces of transparencies the choline to keep within the construct. T he empty coil -construct assemblies made to receive cell -containing beads (the ones with the meshes only) w ere then acid washed and ri nsed to remove proteins left over during
102 manufacturing They were afterwards autoclaved to become sterile and able to receiv e the alginate beads containing a known concentration of tet c ells The alginate beads were generated as described by Simpson et al. (151) A known number of freshly trypsinized cells was mixed with a 2% (w/v) sodium alginate solution u sing alginate with 62% / 38% mannuronic/guluronic content (Novomatrix, Trondheim, Norway) to set an initially cell density in the alginate solution of 3.5 x 107 cells/ml of alginate Aliquots of freshly made beads (~0.3 ml each) were transferred into the construct s using a syringe and a 16 G needle directly after manufacturing The filled construct s w ere then placed in a 5 -cm diameter deep Petri dish containing sterile phosphate buffer saline (PBS, Invitrogen, Carlsbad, CA) for the duration of the NMR ima ging sessions to reduce the TCho NMR signal contamination by glucose NMR signal (see 4.3. 7 for more details on glucose and choline NMR signal interferences ). The PBS was containing antibiotics (100 U/ml penicillin and 100 ng/ml streptomycin) and 20 mM of the buffering agent HEPES (Mediatech, Inc., Manassas, VA) to keep the pH of the culture stable outside of the incubator. PBS was replaced by culture media as soon as the imaging session was done. 4 3 7 NMR Measurements NMR measurements were performed on an 11. 1 T 40 cm clear horizontal bore Magnex magnet equipped with a Bruker Avance console (Bruker Instruments, Billerica, MA). When the receive-only inductively -coupled coil systems were tested, a n in -house custom built transmit only birdcage volume coil provid ed the sample excitation (Figure 4 3 ). Imaging and spectroscopy d ata acquisition and image processing were done using Bruker Paravision software. The spectroscopy data processing was done via a Bruker XWin NMR software. The SNR was determined for each co il system using either a custom image analysis
103 software written in IDL (ITT Visual Information Solutions, Boulder, CO) for images or the XWin -NMR software for spectra. Figure 4 3 Photograph of the RF coil setup to acquire NMR images This setup included a receive-only inductively -coupled coil system and a transmit -only birdcage volume coil. (Photo by N.A. Volland). For coupled -coil system comparisons with phantoms (water and gel) 1H images were acquired using a spin -echo (SE) pulse sequence with a r epetition time (TR) of 1000 ms, an echo time (TE) of 10 ms, 1 -mm slice thickness, 1 average, 6 x 6 cm2 field of -view (FOV) and a 256x256 matrix. PDMS signal was suppressed when required using a spectrally selective saturation pulse centered on the PDMS signa l (5 ppm away from the water signal). For the coupled-coil system comparisons for localized spectroscopy, both localized images and 1H spectra were acquired on a 3x3x3 mm3 voxel using a point resolved spectroscopy (PRESS) sequence (see Section 220.127.116.11) D ifferent pulses were tested for excitation and refocusing to optimize the sequences used for localized imaging and spectroscopy and obtain
104 uniform si gnal intensity across the voxel : sinc3, sinc5, sinc7H, Mao and Mao_4. All of these pulses were based on si nc functions The ir difference was the number of side lobes the sinc function included and the function that multiplied the sinc function to reduce the truncation effect s on the pulse. Sinc3 and Sinc5 pulses were the product of a sinc function with 3 or 5 side lobes respectively and a Gaussian function with a 25% truncation level They were both available in the localized imaging PRESS sequence Sinc7H pulses were the product of a sinc function with 7 side lobes and a Hamming window (162) and available in the localized spectroscopy PRESS sequence The Mao_4 (imaging PRESS sequence) and Mao (spectroscopy PRESS sequence) pulses used an optimized s inc4 function for spin refocusing (163) The voxel imag ing acquisition parameters were a sinc5 excitation and a Mao_4 refocusing pulses, a TE of 10 ms, a TR of 1 s, and 1 average. The localized spectrum acquisition parameters were a sinc7H excitation and a Mao refocusing pulses, a TE of 10 ms, a TR of 2 s and 1 average Adiabatic pulses were applied when localized spectroscopy was performed with the transmit receive inductively-coupled implantable coil systems to alleviate the excitation magnetic field inhomogeneity (164) 1H spectra focusing on the TCho detection were also acquired to assess the sensitivity of the method on identifying this signal peak at 3.2 ppm to later determine the cell viability within the construct. To allow the detection of the Tcho signal a chemical shift selective (CHESS) water suppression schema was used to suppress the water signal (84) It included 3 presaturation sinc3 pulses, one in each spatial direction. Up to 128 averages were used to detect the TCho signal. A n outer -volume -suppression (OVS) schem e was also used when needed to sup press signal coming from outside the voxel and contaminating the NMR signal from the voxel (165) In the OVS schema used here, 3 repeats of a series of 6 presaturation sech pulses were performed
105 on slices outside the voxel and parallel to its surface. Furthermore, care was taken to avoid contamination of the TCho signal resonance at 3.2 ppm by the glucose signal with a resonance at 3.22 ppm Rinsing and incubating the cell containing constructs in PBS prior to and during all the NMR measurements was implemented in this matter 4 3 8 SNR Measurements and Statistical Analysis To assess the perform ances of the different coil s ystems/configurations tested in the magnet, their SNR was evaluated from the NMR images by choosing a signal regionof -interest (ROI) at the IC position and a noise ROI outside of the sample as described in details in S ection 3 .3.8 To assess the performances of the localized spectroscopy method, the SNR was evaluated from the NMR spectra by determining the ratio between the intensity of the TCho signal peak and the average noise intensity from the spectra baseline. The number n, of i ndependently-made coil systems tested to assess the reproducibility of the system construction is indicated for every experiment by (n) At least two set of NMR data were acquired for each coil system to assess the reproducibility of signal detec tion and the sensitivity of the systems when possible The averaged SNR determined for each similar system/configuration was then normalized to the SNR obtained with the TRX system which was used as the reference and normalized to 1. This normalization w as appropriate because the TRX and RX system had the same size and shape. All the results are indicated as a mean standard deviation (relative standard deviations are indicated in parentheses in percentage). A statistical analysis was performed using a two -sample unequal variance t test. The results of this analysis were considered statistically significant when p-values were less than 0.05.
106 4 4 Results 4 4 1 RF Coil System Five different receive -only implantable coils, using three different layouts and two diffe rent diode types, were tested. The three basic layout, shown in Figure 4 4 consisted of 1) loop -gap resonator with one tuning capacitor and Schottky anti -parallel diodes; 2) loopgap resonator with two tuning capacitors and either Schottky anti parallel diodes or two PIN diodes placed in anti parallel configuration, and 3) loop-gap resonator with four tuning capacitors and either Schottky anti -parallel diodes or two PIN diodes placed in anti -parallel configuration. The coil characteristics were tested bef ore and after the introduction of the decoupling circuit to quantify the changes due to the insertion. After the decoupling circuit addition, the coil resonant frequency was shifted and its Q decreased compared to the characteristics of the coil before th e insertion. These changes were attributed to the diode characteristics. The equivalent series resistance of the Schottky anti -parallel diodes (n = 6) was evaluated above 1 their parallel resistance approximately 3 k and their parallel capacitance, at 4.26 0.19 pF ( 4.54%), whereas the coil capacitance varied between 4 and 5 pF when the implantable coil had only one tuning capacitor (Figure 4 4A). When the decoupling circuit was introduced in parallel to the tuning capacitor of the transmit recei ve implantable coil (IC_RX_1M; Figure 44A), its resonant frequency of the coil shifted by over 125 MHz. Furthermore, the diode parallel resistance was low enough to lower the Q of the coil over threefold compared to the transmit receive implantable coil Because of the drastic changes the decoupling circuit introduced in the implantable coil circuit, this layout was quickly abandoned and other designs had to be investigated.
107 Figure 4 4 Photographs of the different receive -only implantable coil layout s built and tested. A) Loop -gap resonator with one tuning capacitor and a decoupling circuit including a hand wound inductor and Schottky anti -parallel diodes; B) loopgap resonator with two tuning capacitors and a decoupling circuit including a hand wound inductor and Schottky anti -parallel diodes; C) loop gap resonator with four tuning capacitors and a decoupling circuit including a hand wound inductor and two PIN diodes. (Photo by N.A. Volland). The first improvement attempted was to double the numbe r of tuning capacitors in the coil to increase their intrinsic value (IC_RX_2M, Figure 4 4 B) and decrease the diode capacitance influence on the coil characteristics. When the receive-only implantable coil comprised two capacitors instead of one, the freq uency shift became only 49 MHz and the Q decreased by only a factor of 2. The number of tuning capacitors in the coil was further increased to 4, the maximum number the coil could practically comprise. Different limiter and PIN diodes with lower juncti on capacitance were also investigated. The PIN diodes (n = 4) were selected for their non -magnetic properties, their parallel capacitance at 1.55 0.07 pF ( 4.56%), their equivalent series resistance below 1 and their parallel resistance approximate 25 k Three more configurations were then characterized: 1) a receive -only implantable coil with 4 capacitors and Schottky anti -parallel diodes (IC_RX_4M); 3) a receive -only implantable coil with 2
108 capacitors and 2 PIN diodes (IC_RX_2C); and 3) a receive -only implantable coil system with 4 capacitors and 2 PIN diodes (IC_RX_4C, Figure 4 4 C). The receive -only implantable coil with 4 capacitors saw its resonan t frequency shifted by 22 MHz, when the decoupling circuit comprising the Schottky anti -parallel diodes was added onto it. Its Q was also lowered by 1.5 compared to same coil without the decoupling circuit. The receive-only implantable coil with 2 capacitors saw its resonan t frequency shifted by 2 5.5 MHz, when the decoupling circuit comprising 2 PI N diodes was added onto it. Its Q was also lowered by 1.9 compared to same coil without the decoupling circuit. The receive-only implantable coil with 4 capacitors saw its resonan t frequency shifted by 6 MHz when the decoupling circuit comprising 2 PIN diodes was added onto it. Its Q was also lowered by 1.13 compared to same coil without the decoupling circuit. These results are summarized in Table 4 1. Even though the Q were determined at different frequencies, the results still indicate an increase in resistive losses as the differences in Q between coils were always larger than their frequency shits Table 4 1. Implantable coil characteristic changes after the addition of a decoupling circuit onto the coil. f represents the coil resonant frequen cy shift and Q the Q ratio between the implantable coil without the decoupling circuit and the same coil after the addition of the decoupling circuit. # of caps indicates the number of capacitors in the coil and # of coils tested the total number of coil tested. Coil # # of coils tested # of caps Diode type f (MHz) Q IC_RX_1M 2 1 Schottky 127.18 7.18 ( 5.64 %) 5.46 3.61 ( 66.11%) IC_RX_2M 4 2 Schottky 49.14 4.38 ( 8.92%) 1.95 0.13 ( 6.69%) IC_RX_4M 2 2 Schottky 2 5.48 0.74 ( 2.91 %) 1.48 0.22 ( 14.88 %) IC_RX_2C 2 4 PIN 22.20 1.27 ( 5.73 %) 1.90 0.08 ( 4.27%) IC_RX_4C 2 4 PIN 6.89 4.11 ( 59.67 %) 1.13 0.07 ( 6.41%)
109 When the transmit receive surface coil received the decoupling circuit in parallel of its fixed tur ning capacitor, the same behavior as that observed on the implantable coil happened. The capacitance of the diodes was large enough compared to the coil capacitors to fundamentally influence the coil resonant frequency (over 100 MHz shift), and lower its Q (approximately by 5). Consequently, the receive -only surface coil was also quickly designed with 2 fixed tuning capacitors instead of one (Figure 4 5A). The frequency shift of the resonant frequency of the coil was then below 30 MHz, and its Q decrea sed by approximately a factor 1.8. When the receive-only surface coil was designed with 3 fixed tuning capacitors and 2 PIN diodes (Figure 4 5 B) to match the best implantable coil design, the resonant frequency shifted was only 3 MHz, and its Q was only lowered by a factor 1.15 compared to the same coil without the decoupling circuit. Figure 4 5. Photographs of the receive -only surface coils tested. A) Two fixed tuning capacitors and a decoupling circuit including a hand wound inductor and Schottky anti -parallel diodes; and B) 3 fixed tuning capacitors and a decoupling circuit including a hand wound inductor and two PIN diodes. (Photo by N.A. Volland). As the receive only inductively -coupled systems with the lowest number of capacitors (one fixed t uning capacitor on each coil) had poor characteristics they were not considered for magnet testing. T wo of the four remaining receive -only inductively-coupled systems constructed were selected to be tested in the magnet. The first system selected was th e system with 2 fixed tuning capacitors and Schottky anti parallel diodes on each coil. This system had the lowest
110 characteristics of the four remaining receive only systems and was chosen to show how the least favorable coil system characteristics influ ence the NMR method sensitivity. The second system was the system with 4 fixed tuning capacitors and 2 PIN anti -parallel diodes on the implantable coil, and 3 fixed tuning capacitors and 2 PIN anti -parallel diodes on the surface coil. This system had the most favorable characteristics of every system built. The coupled -coil system with the Schottky diode (n = 3) has a Q of 35.13 4.22 ( 12.01%), whereas the coupled-coil system with the PIN diodes (n = 1) has a Q of 37.4, when both systems resonate at 470.75 MHz using a 1-cm diameter tube water sample. These results are summarized in Table 4 2. Table 4 2. Receive -only inductively coupled implantable coil system characteristics. Resonan t frequency (f) and quality factor (Q) for the coupled -coil system including the IC with 2 capacitors and a Schottky diode (RX_M) and the system including the IC with 4 capacitors and 2 PIN diode (RX_C). Coil # # of coils tested f (MHz) Q RX_M 3 470.75 35.13 4.22 ( 12.01% ) RX_C 1 470.75 37.4 However, a technical p roblem occurred that prevented the testing of the inductively coupled coil system comprising PIN diodes in the magnet The RF power supply available to run through the surface coil of the receive -only inductively -coupled implantable coil system and turn t he diodes ON at the time of the experiment was not powerful enough. This rendered the tuning of the transmit -only volume coil prior to magnet testing and the magnet testing of the receive-only inductively -coupled coil system with the PIN diodes compromised A DC power supply could have been considered as it was to test the coils on the bench However, it would require the addition of connections on to both coils of the systems This was not applicable to the implantable coil, as it would not be accessibl e when used coated in vitro and in vivo It would have been possible for the surface coil However, leaving the system settings un disturbed while
111 connecting and disconnecting the DC supply was another concern since even small displacement can greatly aff ect the loading and tuning of the systems The tuning and matching of the systems directly in the magnet may have to be investigated in the future along with the use of other diodes with similar favorable characteristics but a lower activation power requ irement to allow the use of these systems in the magnet and improve their characteristics The adequate decoupling performances were also assessed by verifying that there was at least a 20dB difference between the power the coils could transmit to the exte rnal probe when the decoupling circuit was ON compared to when it was OFF. The implantable coils (n = 5) showed a difference of 24.03 0.39 ( 1.64%) dB, while the surface coil showed a difference of 21.8 (n = 1). Only the coils including the Schottky di odes were tested for reasons given in the previous paragraph. 4 4 2 Implantable Coil Coating The hand -wound decoupling-circuit solenoids were found not to be influenced by the PDMS coating. Their deformation during this process was found to be minimal and their characteristics did not change at all. Only one type of implantable coils was coated (the one with 2 fixed capacitors and Schottky diodes) as it was the only functional type. T he coating of these receive-only implantable coils was easily adapted from the transmit -receive implantable coil coating technique and successfully achieved (Figure 4 6 ). The coating thickness on the receive only implantable coil ended up at 1.25 mm instead of the 1 -mm thickness of the transmit receive implantable coil. This was ne cessary to achieve a perfectly sealed coating all around the coil including every component. The frequency shift observed for the transmit -receive implantable coils was also observed for the receive -only coil along with a change in Q. A resonant frequency shift of 5.83 1.62 (
112 27.80%) MHz was detected for the coated implantable coils compared to the non-coated similar coil (n = 4). The quality factor of these coils was no t significantly different from the uncoated coils (within 5% of the non-coated Q) As t his resonant frequency shift d id not create any problems to tune the inductively-coupled coil system ( surface and implantable coils coupled together) at 470.75 MHz, these implantable coils were tested in the magnet F i gure 4 -6 Photograph of a re ceive -only PDMS -coated implantable coil (Photo by N.A. Volland). As loading further changes the coil characteristics, the PDMS -coated coil were embedded in the mouse abdomen like phantom This loading caused an additional resonant frequency shift from t he PDMS coating one of 16.43 1.50 ( 9 .10 % ) MHz, making it impossible to tune the whole system to the desired frequency of 470.75 MHz (n = 3; one coil had an open coating and was remove for the study). This shift in addition to the coating and the de coupling frequency shift were taken into account to build the implantable loop -gap resonator coils which were used for phantom and in vitro studies. However, one more consideration needed to be taken in account here. As the capacitor values were decreased to increase the resonan t frequency of the implantable coil prior to its coating, the resonant frequency shift due to the decoupling circuit
113 addition became larger and larger. The implantable coils to be used loaded were then created with a resonan t freq uency over 30 MHz higher than their predecessors (590.42 5.54 ( 0.94%) MHz) to compensate for this extra shift The implantable coils resonated at 499.08 3.42 ( 0.69%) MHz when coated and embedded 0.74 0.08 ( 11.08%) cm deep in the phantoms (n = 4 ). The quality factor of the loaded implantable coils was affected by the loading and decreased by almost a factor 4 (15.43 2.42 ( 15.68%)) when embedded 0.74 cm in the phantom. Because the diode had such an influence of the tuning and matching of the system the inductively-coupled coil systems could not resonate at the desired frequency of 470.75 MHz after adding the decoupling circuit, coating the implantable coil, and loading it into the gel phantom. The surface coil had to be altered to allow the system to resonate at 470.75 MHz. One fixed tuning capacitance of the surface coil had to be lowered to ma k e the inductively -coupled coil system s resonate at the desired frequency of 470.75 MHz. The Q of the inductively-coupled implantable coil system was dropped down to 9.68 0.78 ( 8.07%) when the PDMS coated implantable coil was embedded 0.74 cm deep in the gel phantom. Once the receive -only inductivelycoupled implantable coil systems were fully characterized, their characteristics could be compar ed to the transmit receive inductively-coupled implantable coil system ones (Table 4 3). If the quality factor of the two sets of implantable and surface coils and the inductively -coupled systems differ significantly when unloaded, the difference fades aw ay when they are loaded. Prior to any in vitro studies, the receiveonly coated implantable coils had to be acid wash and autoclave. These cleaning and sterilization procedure did not alter the coils, which were then ready to receive cell -containing alginate beads.
114 Table 4 3. Comparison of coil characteristics between transmit receive and receive-only coils and coupled -coil systems. Resonan t frequency (f) and quality factor (Q) for both the transmit receive and receive-only implantable coil (IC), surfac e coil (SC), and inductively-coupled implantable coil system (system). Transmit receive system Coil and system type Receive only system n f (MHz) Q f (MHz) Q n 7 49 1 .8 1 3. 38 (0.69%) 203.9 6.20 ( 3.04 %) IC 495.88 3.45 ( 0.70% ) 81.03 12.19 15.04 % 4 4 494.36 2.34 ( 0.47% ) 214.5 20.95 ( 9.77 %) IC_C* 490.12 3.55 ( 0.72% ) 83.49 18.40 ( 22.04% ) 3 2 493.5 72.2 4.95 ( 6.86%) SC 493.5 40.35 2.05 ( 5.08% ) 1 4 49 2 93 3. 44 (0. 70%) 15.41 0.79 ( 5.12%) IC_L# 499.08 3.42 (0.69%) 15.43 2.42 ( 15.68% ) 4 2 493.5 15.20 0.42 ( 2.79% ) SC_L# 493.5 12.05 0.35 ( 2.93% ) 1 7 470.75 96.21 3.70 ( 3.85 %) System 470.75 35.13 4.22 ( 12.01 %) 4 4 470.75 104.82 3.22 ( 3.07% ) System_C* 470.75 30.74 3.31 ( 10.76% ) 3 4 470.75 9.39 0.32 ( 3.7 5 %) System_L# 470.75 9.68 0.78 ( 8.07% ) 4 C indicates that the implantable coils were coated when tested individually and inductively coupled with the surface coil # L indicates that the implantable coil was coated and embedded in a gel phantom when t ested individually and inductively coupled with the surface coil. 4 4 3 Phantom Studies Two different coil -systems/layouts were tested and compared in this study: 1) inductivelycoupled coil system (or TRX system) for both excitation and detection; 2) a receive -only inductively-coupled coil system (RX system) with 2 fixed tuning capacitors and Schottky anti parallel diode s on each coil and a bird cage volume coil for excitation. Figure 4 7 illustrates differences in SNR for these two different types of coils whe n studied under unloaded conditions (water sample) at 11.1 T The se systems were also tested with their implantable loop -gap resonator s coated with a PDMS layer. Their SNR resul ts were compared to each other using the SNR obtained with the TRX system. T he SNR determined from the images of the water sample (equivalent to unloaded conditions) acquired with the receive-only inductively -coupled
115 implantable coil systems was never more than 25% below the one when the images were acquired with the inductively -c oupled implantable coil system. The RX system shows a 23% SNR decrease compared to that of the TRX system [0.77 0.18 ( 23.52%), p << 0.001] when the IC was placed approximately 1 cm away from the SC [0.87 0.08 ( 9.66%) cm] H owever, the PDMS coatin g of the implantable coil makes the SNR of the RX system higher than the SNR of the TRX system with a PDMS coated implantable coil by approximately 10% [0.97 0.22 ( 22.51% ), p << 0.001] The PDMS suppression scheme used to acquire the images did not si gnificantly alter these previous result [0.94 0.20 ( 21.82% )]. Figure 4 7. SNR comparison between transmit -receive and receiveonly inductively-coupled implantable coil systems under unloaded conditions using a water sample TRX system represents t he transmit receive coupled -coil system and RX system the receive-only coupled -coil system where the IC was placed 1 cm away from the SC C indicates that the IC was PDMS -coated. PDMSsup indicates that PDMS suppression was used during the NMR data acquis ition. denotes the statistically difference between the SNR of the transmit receive coupled -coil and the receive -only systems; the difference between the SNR of the receive -only coupled -coil system with an IC, coated or not.
116 The effect of loading on t he measured SNR for both coil systems was also investigated. Figure 4 8 depicts the SNR obtained at the IC (macroconstruct) location under loaded coil system conditions (0.74 cm) using either the RX system or the TRX system with both a PDMS coated loop ga p IC with and without PDMS suppression scheme These measurements were also compared to the SNR ob tained with a loaded TRX system The RX system did actually perform better than the TRX system under loading condition. It consistently displayed a SNR 13% higher than the TRX IC system [1.13 0.19 ( 16.71%), p << 0.001] U sing a PDMS suppression scheme when acquiring the NMR images with coated coils did not significantly affect this SNR [1.10 0.19 (16.96%)] Figure 4 8 SNR comparison between tra nsmit -receive and receiveonly inductively-coupled implantable coil systems under loaded conditions using a gel phantom. TRX system represents the transmit receive coupled -coil system and RX system the receive -only coupled coil system where the IC was pla ced 1 cm away from the SC. C indicates that the IC was PDMS -coated. PDMSsup indicates that PDMS suppression was used during the NMR data acquisition. denotes the statistically difference between the SNR of the transmit receive coupled -coil and the rec eive -only systems.
117 Furthermore, the magnetic field distribution generated by volume coil made the RX system much more suitable for localized spectroscopy as it is more uniform than the magnetic filed distribution generated by the TRX system. Figure 4 9 sh ows the response profile obtained with the transmit receive (TRX) and the receive-only inductively -coupled (RX) systems. The profile differences are clearly displayed. The response profile obtained when using the RX system was much more homogeneous across the construct (within the implantable coil) in any direction compared to the response profile from the TRX system. 4 4 4 NMR Spectroscopy S tudies 1H spectra were acquired without using any volume selective methods on all the solutions with different concentrati ons (100 mM, 10 mM, and 1mM). The 3.2 ppm choline resonance was easily identified on the 100 mM solution spectra since it was even observable when no water suppression scheme was used. However, water suppression was necessary to observe the choline resonance at 3.2 ppm for the less concentrated solutions. Since physiological concentration of choline is closer to 1 mM, 1H localized spectroscopy was then performed on a 1mM choline/glycerin sample (unloaded conditions) using a PRESS pulse sequence on a 3x3 x3 mm3 cubic voxel entirely confined within the construct. Either coupled coil systems, transmit -receive and receiveonly, were used. Images of the voxel of interest were acquired first to demonstrate the feasibility of the volume selection and the local ization of signal within the construct (Figure 4 10). Volume selection was achieved with both RF systems since only signal from within the choline sample can be detected on Figure 410B and Figure 4 10E.
118 Figure 4 9 Signal profiles obtained from NMR im ages acquired with TRX and RX inductively coupled systems A TRX system was used to acquire image A and D, while a RX system was used to acquire image G and J. Coronal images of a water sample are presented in A and G, while transversal images are presen ted in D and J. S ignal profile s from the readout direction of image s A, D, G, and G are presented in B, E, H, and K respectively. Signal profile s from the phase encode direction of image s A, D, G, and G are presented in C, F, I, and L. The yellow circle s and lines represent the coils of the different systems used to acquire the images. The circle in A and G represent the IC, as well as the two short horizontal lines in D and J. The long vertical line in D and J shows the position of the SC below the sam ple.
119 Optimization of the m ethod was explored to obtain the cubic voxel originally selected with the desired size and a more uniform signal intensity across this voxel. The Sinc5 pulses for excitation and the Mao_4 pulses for refocusing were found to be the most appropriate pulses available to use in the PRESS imaging sequence Signal could be detected from outside the voxel. While an OVS scheme could have helped suppress these undesired signals, the OVS was not available when the volume selective imaging sequence was used to locate the voxel of interest. Figure 4 10. Volume selective imaging acquisition in preparation for 1H localized spectroscopy A transmit receive inductively -coupled implantable coil system was used in A through C A receive -only inductively -coupled implantable coil system was used in D through F. A & D represent scout c oronal image of a glycerin vial including 1 -mM choline sample at its center B & E represent the voxel v olume selective image C & F represent the subtraction o f image s B & E from image s A & D respectively. The white square on A & D represents the voxel selected. The white dashed circle in B & E represents the outer edge of the glycerin vial
120 1H spectra were acquired next and choline signal was detectable from the 1mM choline solution with a SNR of 12.5 after the water signal was suppressed from the spectrum (Figure 4 1 1 C). OVS schema was used to decrease the glycerin signal detected from outside the voxel, al though the signal could not be completely removed. Better shimming procedure and pulses selection should alleviate this issue later on (see S ection 4.5). Choline was also detected from a 1 mM choline solution within an altered coil -construct assembly soaked in PBS (loaded condition, Figure 4 1 1 E) with a S NR of 16.41. Figure 4 1 1 1H localized spectroscopy data acquired using a transmit receive inductively coupled implantable coil system A 1mM choline [Cho] sample surrounded by a glycerin [Gly] solution in a 1cm NMR tube was used in A through C and i n an IC construct assembly soaked in PBS in D and E A & D represent spectra acquired without water suppression or outer volume suppression. B & E represent spectra acquired with water suppression [(H2O)] but without outer volume suppression [Gly] C rep resents a spectrum acquired with both water suppression [(H2O)] and outer volume suppression [(Gly)] The data were acquired using a PRESS sequence with a 1ms sinc7H excitation pulse and a 1ms Mao refocusing pulse, TE = 10 ms, TR = 2 s and 128 average. A diabatic pulses were used.
121 No line broadening was applied during the processing of these data. Even though the volume selective method was not fully optimized, these spectra show the potential of the use of inductively-coupled coil systems with NMR method s and the feasibility of the NMR spectroscopy methods. No s pectrum was acquired using the receive -only inductively-coupled implantable coil system on the choline glycerin sample or the choline sample loaded in the phantom gel. Cells were successfully enca psulated in alginate beads and inserted in the implantable coil construct cavity of both types of coil 1H localized spectroscopy was attempted with the transmit receive implantable coupled -coil system T he parameters used with the 1mM choline solution w ere utilized in this experiment as well Choline was never detected However, the viability of the cells was questionable T he nu m ber of cells in the sample may have been too low and will need to be increase d in future experiments, at least initially to optimize the method and determine its limit of sensitivity 4 5 Discussion NMR spectroscopy (MRS) is a powerful technique to obtain metabolically and physiologically relevant compound information from a sample. D etecting the total choline signal from within the bioartificial pancreatic substitute using localized spectroscopy would allow a direct assessment of the cell viability in the construct (5,6) However, the proper localization of the spectroscopic data is often crucial for the accurate determina tion of the studied compound characteristics at a microscopic level (choline concentration) and for its link to an important macroscopic state (cell viability). To facilitate MRS data acquisition within a bioartificial construct and enhance the method vol ume selectivity, the use of a receive only inductively-coupled implantable coil system for MRS was explored. This coupled -coil system ha s substantive advantages over a transmit -receive inductively-coupled implantable coil system
122 for the study of tissue -en gineered constructs because of the more uniform NMR excitation produce by a volume coil : 1) the measured NMR signal is more homogeneous throughout the construct (Figure 4 9) 2) the volume selectivity of the MRS techniques is enhanced and easier to impleme nt, and 3) the power requirements are not so demanding since the use of adiabatic pulses is not required. A homogenous excitation in MRS is key to obtain ing accurate information from an implanted construct. The receive -only inductively -coupled implantable coil system showed a more homogeneous response NMR signal than the transmit -receive inductively-coupled implantable coil system indicating a better B1 excitation field distribution. Initially, t he sensitivity (SNR) of the two systems g ave advantage to th e transmit receive inductively-coupled system under unloaded condition (Figure 4 7) This c ould be explained by the drop in Q of the receive-only system compared the transmit receive system by a factor of 2.75 (from 96.21 3.70 ( 3.85%) to 35.13 4.22 ( 12.01%). The addition of more components in the receive -only circuit of both coils particularly the low Q diode, is the reason of the lower Q of the receive only coupled coil system However, the SNR only dropped by a factor 1.30 (0.31 ( 23.52%)) wh ich may be attributed to the higher average NMR signal when a receive -only coil system is used compared to when a transmit receive coil system is used because of the higher homogeneity of the signal throughout the sample To alleviate the frequency shift and the Q and SNR decrease, diodes with smaller parallel capacitances higher Q and low power activation requirement could be used. T he diodes found commercially thus far with such favorable characteristics were magnetic, such as the SM1330 limiter diode s (Skyworks, Woburn, MA). However, under loaded conditions the SNR difference between the two systems is minimal, down to 13% and inverted, the receive only system performed better than the transmit -receive
123 system (Figure 4 8) This can be explained by t he fact that the loading effects are large on both coil systems especially at the high frequency they performed at (470.75 MHz) that their Q are very similar, and that the average signal obtained using the receive -only system is higher than that obtained with the transmit receive system. In addition to t he higher detected signal homogeneity when the receive-only inductively -coupled implantable coil system was used, these results are encouraging and showed the potential of the use of such receive -only coupled coil systems to monitor a bioartificial pancreas non invasively using NMR imaging and spectroscopy methods N oise was added into the image when the receive -only inductively coupled coil system was used. This noise was signal detectable from the sample outside of the construct due to the volume excitation even outside of the construct. This was n ever seen wh en a transmit -receive inductively-coupled implantable coil system was used (more obvious in the coronal plan, construct cross -section) as the NMR e xcitation was mainly localized inside the construct. However, this was not an issue and as this noise was located far from the region of interest within the construct. The fact that the receive only inductively -coupled implantable coil system with the PD MS -coated implantable coil tested under unloaded condition gives an SNR higher than its uncoated counterpart and the transmit receive inductively coupled implantable coil system with a PDMS -coated implantable coil has no founded explanation. It is not eve n consistent with the Q difference between all these systems as this difference is largely unfavorable to the receive -only inductively -coupled implantable coil system with the PDMS coated implantable coil. The PDMS coating procedure for the implantable coi l was easy even with the additional components included in the coil to coat. The coating acted similarly on the receive -only coil as it
124 did on the transmit receive implantable coil. T he coating thickness was increased to 1.25 mm to optimize the coating o f all the component of the receive-only implantable coil without compromising the construct space and the system detection sensitivity when loaded. T he outer space taken by the coil increased due to the additional components (use of a larger hollow punch) The frequency shift introduced by this coating along with the one introduced by the decoupling circuit was taken into account when building these coils. T he implantable coil was made to resonate as closely as possible to 590 MHz to allow the receive -on ly inductivelycoupled system to tune and match at 470 .75 MHz when the implantable coil was coated with PDMS and embedded with the gel phantom or used in vi tr o If the resonan t frequency shift and Q change of the receive -only implantable coil when coated and loaded were not as pronounced as when a transmit receive implantable coil was coated and loaded it was changing with the coil capacitors values. The frequency shift due to the decoupling circuit addition, the coating and subsequently the loading w as getting larger and larger as the capacitors values were decreasing to increase the coil initial resonant frequency to try to compensate for the shifts due to the coating and loading. Similarly, the Q also decreased further and further with the decoupl ing circuit addition, the coating and the loading, as the capacitors values were decreased to make the coil resonate at a higher and higher frequency to compensate for all the shifts Consequently, t he larger capacitor values used for the receive-only i mplantable coil construction compared to the ones used for transmit -receive one may make the effects of coating and loading on the coil less influential. Nonetheless, the decoupling circuit addition created a lot more discrepancy (larger deviation of the results) in the coating and loading results than that found for a transmit receive coil. Also, at least one receiveonly implantable coil stopped working properly after every set of tests showing how fragile these coils have become due to their larger siz e and
125 increased number of components. Even a careful handling c ould not insure coil integrity for an extended period of time. The larger the coil surface is, the more likely the coil is subject to deform and its functioning to be impaired. This c ould be a problem with possible deformation of the implantable coil -construct assembly at the time of surgical implantation and also over time when floating in the peritoneal cavity of a recipient. This issue will have to be inves ti gated more in depth when in vivo experience will be preformed. However this observation brought the issue of the implantable -coil construct assembly size limitation for the animal model (mouse) used in this study with the use of a slightly larger implantable coil of. Even with the m ost careful construction, the smallest receive -only implantable coil with the characteristics presented here cannot fit in a mouse peritoneal cavity anymore. Moving to a bigger animal model such as rat is one possibility. Using miniaturized integrated ci rcuits is another possibility. If not explored here, there are certainly to be investigated in the future to make the coils more stable, their construction more consistent, and their use possible in vivo Localized spectroscopy was also performed using both inductively -coupled coil systems to assess the superior ability of the receive -only inductively -coupled coil system in performing localized spectroscopy. Choline signal was detected from a 1 mM solution under unloaded and loaded conditions when a trans mit receive inductively -coupled implantable coil system was used while the technique was not optimized. This demonstrate s the potential of performing localized spectroscopy with inductively-coupled implantable coil system. With the improvement of the sh imming procedure and a better selection of the pulses used, in vitro and in vivo studies will be possible Often only the linear shims were used to make the static magnetic field as homogenous as possible within the voxel. If signal line width of 5 to 15 Hz could be obtained under unloaded conditions, it was never achievable under loaded conditions. Manual shim and
126 fast automatic shimming technique by mapping along projection (FASTMAP, (166) ) were explored occa sionally. However, a FASTMAP calibration for the excitation and detection system and the voxel size required in the study may help in obtaining more optimal results in the future Better water suppression and OVS could also benefit from a better shimming over the voxel of interest. Improving the RF pulse profile s to obtain a better voxel shape and signal uniformity across the voxel would be the second step toward a better localized spectroscopy. The RF_profile sequences included in the Paravision softwa re were mainly made for volume coil excitation and detection as the profile can only be obtained in the direction of the static magnetic field and not in any other direction as needed for the system s investigated in this study. One type of PRESS sequences (t he one designated as ParaVision Method or PVM ) had some profiling capability to be explored once the shim procedure is optimized Moreover, with increasing magnetic field, longitudinal relaxation times (T1) tend to increase and transversal relaxation t imes (T2) tend to decrease (152) No previous work was found to lead to a T2 range for total choline at 11.1T. However, a stimulated echo acquisition m ode (STEAM see Section 18.104.22.168) pulse sequence may be used to acquir e these data if total choline T2 is found to be relatively short (83) Th ese are suggestions to consider in the future to enable the optimum de tection of the total choline NMR signal at high field (11.1T and up) to monitor the functioning of a bioartificial pancreatic construct with cell viability. Furthermore, the buffering agent HEPES used to maintain a physiologic pH for the cells in PBS outsi de of the incubator (when CO2 is in limited quantity) when NMR experiments were conducted in vitro was found to have a resonance close enough to the 3.2 ppm of the choline and could have potentially mask ed it. It was subsequently removed of the PBS soluti on already used instead of culture media to eliminate the glucose interference with the choline resonance at 3.2
127 ppm. Furthermore, the fact also the no choline signal could be detected from the in vitro sample with cells brought the issue of the viability of the cells in pure PBS for an extended period of time. The period of time during which the cells can be kept alive in PBS at room temperature will have to be attested before new experiments are conducted. 4 6 Summary Receive -only i mplantable inductively -co upled loop-gap coils assembled within bioartificial pancreas macroconstructs have been constructed and their performance analyzed on the bench, in phantoms, and in vitro NMR images from coil construct assemblies were obtained under unloaded and loaded conditions in gel phantoms at 11.1 T, demonstrating the capability of a receive -only inductively-coupled implanted coil system to perform NMR imaging and spectroscopy to monitor a bioartificial pancreas non -invasively T he SNR gain that th e receiveonly inductively coupled implantable coil system has over a transmit receive system at 11.1 T was similar under loaded conditions ( in vitro and in vivo environment equivalent) However t he increased excitation homogeneity and selectivity available with the use o f this receive -only inductively-coupled implantable coil system allows further gains in information obtained from an implanted construct Spectra were acquired using a transmit -receive inductively-coupled coil system and t he Cho signal could be observed fr om the construct location even though the localized spectroscopy method was not fully optimized These results show the potential of performing localized spectroscopy with high sensitivity us ing inductively-coupled implantable coil systems to detect the TCho signal from live cells and link this information to cell viability within the construct. This give s a fundamental basis toward the detection of early marker of pancreatic construct failure in vivo
128 With the feasibility assessment of receive-only indu ctively -coupled implantable coil systems, the focus of this research was next directed toward the feasibility assessment of another type of complex inductively -coupled implantable coil system: multi -frequency system s These multiple -frequency coupled -coil system s are of primary interest as they would allow the detection of less-sensitive, yet important nuclei such as 31P and 19F, and generate direct correlation of the results from the different nuclei detection to fully characterize a bioartificial constr uct
129 CHAPTER 5 TOWARD THE DEVELOPME NT OF A MULTIPLE -FREQUENCY INDUCTIVEL Y COUPLED RF COIL SYST EM FOR SIMULTANEOUS 1H, 19F, AND 31P DETECTION FOR IMAGING AND SPECTROS COPIC ANALYSIS OF AN IMPLANTABLE BIOARTIF ICIAL CONSTRUCT AT 11.1T 5 1 Introduction Monitoring the viabi lity and the bioenergetic s status of cells as well as the oxygen concentration within a bioartificial pancreas, is of primary importance to directly reflect the well being and functioning of the construct. NMR imaging (MRI) and spectroscopy (MRS) have sho wn to be well -suited techniques for such a non -invasive and temporal monitoring of such a device in vitro and in vivo However, to achieve a complete monitoring, these NMR techniques depend on the detection of specific nuclei including, but not limited to hydrogen (1H), phosphorus (31P) and fluorine (19F) In this study, th e detection of physiological levels of these nuclei wa s investigated while using high sensitivity single -frequency inductively -coupled implantable RF coil systems at high field (11.1T ). Moreover, the feasibility of multiple frequency inductivelycoupled implantable coil systems, (a 1H -31P & a 1H -19F one ), w as examined as only one implantable coil can be implanted with the construct at a time. Standards were established using single -f requency inductively -coupled implantable coil systems and the feasibility of double frequency inductively -coupled implantable coil systems was assesed 5 2 Background The well being and functioning of a bioartificial pancreas can be directly assessed by the mo nitoring of the viability and the bioenergetic s status of its cells as well as the oxygen concentration within the tissue -engineered construct. In order to perform such a complete monitoring of a bioartificial pancreas using NMR methods t he nuclei of interest include, but are not limited to, hydrogen (1H) (5,6) phosphorus (31P) (47,91,95) and fluorine (19F) (94 96) To detect these different nuclei different NMR methods can be developed. However, due to the low
130 NMR sensitivity of some of these nuclei the methods must have an optimal detection sensitivity to be able to monitor physiological level of these nuclei. If the method developed by Stabler et al. showed promises in detecting 1H choline signal and linking it to the cell viability within the construct (5,6) this method sensitivity threshold prevented its use with low densities of cell within bioartificial construct and the detection of other less sensitive nuclei such as 31P. Recently, the use of higher magnetic field strength along with inductively-coupled implantable coil systems was shown to improve the sensitivity of th e NMR method s by a f actor of 4 (160) This would allow the detection and study of lower density of cells within constructs in vitro and in vivo as well as less sensitive nuclei (compared to 1H), such as 19F and 31P However, only one coil can be implanted per construct Furthermore, th e construct is neither retrievable from the middle of the coil ( without being impaired ) nor interchangeable T his shows the need for a combined coil system tuned to multiple frequencies that would gather information from more than one nu cleus at once and allow an easy correlation of the results obtained for the different nucleus signal simultaneous detection For in vivo studies, these multiple -frequency coils are generally birdcage volume coils (112) or surface coils (113,114) essentially double -tuned 1H and 31P, due to the importance of the biological compounds observed a t these frequencies (112,113,115,116) These RF coils are mainly based on tank, or trap, circuits (107) and transformer -coupled circuits (108) Tank circuits are introduced in to the circuits to add poles in the complex reactance equation and create additional resonan t frequencies. T ransformer -coupled (or inductive overcoupling) method use s inductive coupling to introduce additional resonance in the c ircuits. The efficiency of these two methods is relatively similar (113) However, transformer -coupled coils do not seem appropriate for the purpose of this study since inductive overcoupling is already used to enhance the
131 sensitivity of the NMR methods under investigation (see S ection 3.3.3 for inductively-couple implantable coil system ) focusing on only one of the resonan t frequencies of the system (the co rotating mode) Furthermore, the distance between the implantable and surface coils is fixed by the design of the experiment studied an d the anatomy of the implantable -coil recipient such that the degree of coupling may be too small for the successful operation of the transformer -coupled double -f requency coil system. Thus, t his report investigates and discusses the feasibility, the techn ical aspects and the use of multiple -frequency (1H -31P and 1H -19F) inductively-coupled RF coil systems using tank circuits at a high field (11.1 T) to monitor implanted tissue -engineered constructs with NMR It also specifies the maximum performances the se types of systems will achieve for 19F and 31P detection by testing single -frequency inductively coupled implant able coil s ystem (one 19F and one 31P system) detecting: 1) the -P signal of adenosine triphosphate (ATP), which can be linked to the cell bi oenergetics within the bioartificial pancreas ( BAP ), and 2) the 19F signal of added perfluorocarbons which can be linked to the oxygen tension within a construct. 1H spectroscopy and the choline signal link to the cell density within a BAP was investigat ed in Chapter 4. 5 3 Materials and Methods 5 3 1 RF Coil S ystem D evelopment Single -frequency and multiple -frequency i nductively-coupled implantable coil systems were constructed tested and compared at 11.1 T Following the results obtained in Chapter 3, t h e five coupled -coil system s developed in this study consist ed of an implantable loop-gap resonator coil inductively coupled to a surface coil for NMR imaging and spectroscopy. All systems function as both transmitter and receiver at every frequency. The 1H and 19F single frequency inductively coupled implantable coil systems were similar to each other and similar to
132 the system described in S ection 3.3.1 (160) T he implantable coil bioartificial pancreatic macroconstruct assembly was s imilar to the one described in S ection 22.214.171.124. Only the fixed capacitors values were different to reflect the difference in resonan t frequency of 1H and 19F. The single -frequency inductively -coupled implantable coil system had to be slightly modified to obtain a practical 31P single -frequency coupled -coil system. Since 31P has such a low NMR sensitivity (6% sensitivity compared to 1H), sample positioning at the center of the magnet and shimming procedure are usually done by detecting a more sensitive nuclei such as 1H. Consequently, t his 31P system had a double frequency (1H -31P) surface coil (Figure 5 1 A ) and an implantable coil (similar to the one presented on Figure 3 2A) only tuned to 31P frequency. The surface coil was not optimized to 1H resonant frequency (not matched to 50 ) as it was only used for posi tioning and shim purposes. To render the two last coupled -coil system s double -frequency the addition of tank circuit s on both the implantable and surface coils of the single -frequency system was required (107) in order to introduce extra resonance to the circuit (Figure 5 1). A tank circuit consists of a parallel capacitor and inductor added in series with a tuning capacitor of a resonant circuit. The tank circuit is a resonant circuit in itself. It is inductive at the lower circuit resonan t frequency since the resonant frequency of the tank circuit is higher than the main circuit lower resonant frequency. At the higher circuit resonant frequency, the tank circuit becomes more capacitive as the resonant frequenc y of the tank circuit is lower than the main circuit resonan t frequency This allows the tuning of both resonan t frequencies individually (118) To generate the inductively coupled double -frequency 1H -31P implantable coil s ystem, a standard tank circuit was added onto the implantable and surface coil (190.5 MHz and 470.75 MHz were the respective resonan t frequency of 1H and 31P at 11.1 T ). The double -frequency
133 implantable coil was developed based on the design of the loop-g ap single -frequency implantable coil. Its inductor was made of a 2 -mm-wide 202m thick copper foil and had a 1.2 -cm diameter (Li on Figure 5 1 A ). Its capacitive part included two fixed tuning capacitors (CT ui, CT ai on Figure 5 1A) which divided the loo p into two parts to spread the voltage more evenly throughout the circuit and allow the addition of an inductor in parallel to one of the capacitors to form a tank circuit. The double -frequency surface coil inductor was also constructed based on the chara cteristics of the surface coil of the single -frequency system. It was a single 2 -cm diameter circular loop of 2 -mm wide 35 m thick copper tape ( Ls on Figure 5 1) held flat on a 5mm thick Plexiglas platform. The capacitive portion of the surface coil incl udes two variable matching capacitors ( Voltronics Corporation, Denville, NJ 1 15 p F CM s 1 V and CM s 2 V on Figure 5 1 B), one fixed tuning capacitor ( American Technical Ceramics, Hartford, CT CT us on Figure 5 1 B), one variable tuning capacitor ( 1 15 p F, CT us V on Figure 5 1), and one variable tank capacitor ( 1 15 pF, CT a s V on Figure 5 1 B). The fixed tuning capacitor placed on the inductive loop divides this loop into two parts to distribute the voltage more evenly throughout the circuit. The addition of an i nductor in parallel to the variable tank capacitor formed the tank circuit. T he tank circuit and three variable capacitors were soldered onto a circuit board to provide variable tuning and matching of the system to 50 at both frequencies A coaxial cable (RG 58 C/U) connected the coil to either the 1H source and preamplifier or the 31P ones. Two cable traps were added to the shield of the coaxial cable to block shield currents that contribute to parasitic losses whe n working at both these high frequency (140) The 31P cable trap was pla ced at the circuit board end of the coaxial cable and the 1H a wavelength away from the circuit board end. As p hosphorus is the least sensitive nucle us to be studied here
134 the design of the system w as assembled in a way that phosphorus was optimized ove r the other nuclei. Figure 5 1 Circuit diagram for an inductively-coupled double -frequency (1H -31P) implantable coil system. A ) An implantable coil; and B) a surface coil. Inductors (L) and capacitors (C) are shown with subscripts denoting implantable (i) or surface (s ) coils Capacitors are also designated as tuning (CTu), tank (CTa), or matching (CM). M indicates the mutual inductance or inductive coupling and V variable. A cable trap equivalent, including an inductor and a capacitor, is also sh own in the diagram. T he 19F -1H double -frequency inductively -coupled coil system was implemented slightly differently than the 31P -1H system since 19F and 1H resonated too close to each other to make the system work with a tank circuit only. T he fact that 19F and 1H have resonant frequenc ies close to each other (442.86 MHz and 470.75 MHz respectively at 11.1 T ) require d circuit modifications to avoid large resistive losses at both frequencies This can be explained by the l arge current that will be circula ting in the tank circuit because the tank circuit resonant frequency will be close to both resonant frequencies To allow one circuit to resonate properly at multiple frequencies when frequencies were close to one another such as the case of 19F and 1H, c apacitive coupling (120,167) was used between the tank circuit and the main circuit To generate the inductively coupled d ouble -frequency 1H -19F implantable coil system, a tank circuit was capacitively
135 coupled to both the implantable and surface coil. The implantable coil of this double -frequency system was developed based on the design of the loop-gap single -frequency impla ntable coil. Its inductor was a 1.2-cm diameter 2 -mm-wide 202m -thick copper foil (Li on Figure 5 2 ). A fixed tuning capacitor was soldered between the ends of this resonator loop (CT ui on Figure 5 2 A ). A tank circuit including a capacitor (CT ai on Figure 5 2 A ) and an inductor (LT ai on Figure 5 2 A ) was added in parallel to this tuning capacitor along with a coupling capacitor (CKi on Figure 5 2 A ). The surface coil inductor of this double -frequency system was also constructed based on the characteristic s of the single -frequency surface coil. It was a single 2 -cm -diameter circular loop of 2-mm wide 35 m -thick copper tape ( Ls on Figure 5 2) held flat on a 5mm thick Plexiglas platform. The capacitive portion of the surface coil includes two variable match ing capacitors ( Voltronics Corporation, Denville, NJ 1 15 pf, CM s 1 V and CM s 2 V), one fixed tuning capacitor ( American Technical Ceramics, Hartford, CT CT us on Figure 5 2 B ), one variable tuning capacitor ( CT usV on Figure 5 2 B, 1 15 pf ) in parallel to a tan k circuit including a variable capacitor (CT a sV on Figure 5 2 B) and an inductor (LT a s on Figure 5 2 B) along with a variable coupling capacitor (CK s on Figure 5 2 B). The fixed tuning capacitor was placed on the inductive loop and divided it into two parts to spread the voltage more evenly throughout the circuit. The tank circuit, the tuning and the two matching variable capacitors were soldered onto a circuit board to provide variable tuning and matching of the system to 50 at both frequencies A coaxial cable (RG 58 C/U) connected the coil to either the 1H source and preamplifier or the 19F ones. Two cable traps were added to the shield of the coaxial cable to block shield currents that contribute to parasitic coupling w hen working at both these high frequency (140) The 19F cable trap was p laced at the circuit board end of the coaxial cable and the 1H a wavelength away
136 from the 19F one. As Fluorine is a less sensitive nucleus than 1H, the design of the system w as assembled without trying to optimize one nucleus over the other one. Figu re 5 2 Circuit diagram for an inductively-coupled double -frequency (1H -19F) implantable coil system. A ) A n implantable coil; and B) a surface coil. Inductors (L) and capacitors (C) are shown with subscripts denoting implantable (i) or surface (s ) coil Capacitors are also designated as tuning (CTu), tank (CTa), or matching (CM). M indicates the mutual inductance or inductive coupling and V variable A cable trap equivalent, including an inductor and a capacitor, is also shown in the diagram. Even t hough the multiple -frequency coupled -coil system requires more components than the single frequency system, the size of the implanted coil needed to be kept small to be usable in vivo This size restriction also applie d to the surface coil as tank circuits are resonating loops themselves and can introduce undesired loss in the main circuit if made too large. The inductors of the tank circuits ( LTa s and LTai on Figure 5 1 and 5 2) were small solenoids made of 25 AWG (American Wire Gauge) copper wires hand wound around a 16 AWG copper wire making their self resonance much higher than the tank and the main circuit resonance. Their values were verified using the RF impedance test adapter HP 1620A spring clip fixture of the network analyzer ( 8752C Hewlett Pack ard Santa Rosa, CA) These inductors were placed perpendicular to the main coil loop of the implantable coil to limit the interaction between each other ( LTai on
137 Figure 5 1 A and 5 2 A ) and over the tank capacitor (LTa s and CTa sV on Figure 5 1 B and 5 2B ) o n the surface coil circuit board This limited signal degradation due to the addition of extra undesired frequencies in the system because of inductively coupling between the tank inductor and the main circuit inductor. Furthermore, t o allow further flex ibility in the frequencies tuning and the matching to 50 of both double frequency coil systems the capacitor of the tank circuit (CTa sV on Figure 5 1) and the coupling capacitor of the surface coil s (CK sV on Figure 5 1) were a lso variable. 5 3 2 Coating Techn ique s The material used for th e coating of the implantable coil s used in this study was the same one as the one used for the transmit receive and receive -only implantable coil presented in Chapter 3 and 4: polydimethylsiloxane (PDMS, Medical Grade Silicone Elastomer, Factor II, Lakeside, AZ) The technique applied has been described in details in S ection 126.96.36.199 (160) However, a daptations w ere made to th is coating procedure to accommodate th e double frequency coil extra components. The coating of the hand -wound tank -circuit solenoids was tested prior to the coating of the entire coil as described in S ection 4.3.2 as they could be easily deformed under pressure resulting in drastic cha nges in their characteristics (inductance in this case). 5 3 3 RF Coil Testing When surface and implantable coils are in proper orientation and close proximity, they inductively couple (143) as described in S ection 3.3.3. In order to have the 19F single -frequency coil system resonating at 442.86 MHz, the implantable and surface coil had to be individually tuned and matched to a h igher frequency ( 464.8 MHz) i n the (+) mode and to a lower frequency (421.6 MHz) in the ( ) mode In order to have the 31P single -frequency coil system resonating at
138 190 MHz, the implantable and surface coil had to be individually tuned and matched to a h igher frequency ( 199.6 MHz) i n the (+) mode and to a lower frequency ( 181.9 MHz) in the ( ) mode T he inductive coupling o f the surface and implantable coils of the two double -frequency inductively-coupled implantable coil systems induced a total of 4 res onant frequencies since every coil already had 2 resonant frequencies each. This means that the systems had 2 current modes for each resonant frequency In order to have the double -frequency 31P -1H inductively coupled coil system resonating at both 190.5 MHz and 470.75 MHz, the implantable and surface coil had to be individually tuned and matched to a higher frequency for both frequency (19 5 8 and 475.5 MHz) in the (+) mode, whereas the implantable and surface coil had to be individually tuned and matched to a lower frequency for both frequency (185.4 and 465.8 MHz) i n the ( -) mode In order to have the double -frequency 19F -1H inductively-coupled implantable coil system resonating at both 442.86 and 470.75 MHz, the implantable and surface coil had to be i ndividually tuned and matched to a higher frequency (449.3 and 481.8 MHz) i n the (+) mode whereas the implantable and surface coil had to be individually tuned and matched to a lower frequency for both frequency (436.4 and 459.7 MHz) i n the ( -) mode T he five different inductively -coupled systems developed were: 1) one single -frequency system for 1H detection (470 MHz) (referred as to 1H system); 2) one single -frequency system for 19F detection (442.86 MHz); 3) one single -frequency system for 31P detection (190.5 MHz); 4) one double -frequency system for both 1H and 31P detection; and 5) one double -frequency system for both 1H and 19F detection. For all these systems all variable capacitors CM 1 V, CM 2 V, CTusV, CT asV, and CKs V ( Figure 5 1 and 52) in the sur face coil s were adjusted to achieve a match to 50 at each resonan t frequency. These differ ent systems were first simulated with antenna analysis software (Graphical Numerical Electromagnetics Code, Nittany, Inc. Riverton, UT).
139 These systems were then o ptimized on the bench with a network analyzer (Hewlett Packard 8752C, Santa Rosa, CA) before testing in the magnet. The quality factor (Q) of each coil taken individually and of the whole system was recorded after construction with and without load and ev ery time the system was tested in the magnet. The measurements were performed as described in details in S ection 3.3.3. A coupling factor was also determined for every coupled -coil system to characterize its inductive coupling. Usually, t h is coupling fac tor ( k) represents the amount of magnetic flux transmitted from one coil (the surface coil in the case of transmission in this study for example) to the other (the implantable coil here) and is often described as: SC IC.L L M k ; (5 1) w here k is the coupling factor, M the mutual inductance of the inductively-coupled implantable coil system, LIC the total inductance of the implantable coil of the system, and LSC the total inductance of the surface coil of the system (118,142) However, the determination of the coupled coil system mutual inductance is n ot eas y or exact (168) Here, the coupling factor was determined for every coupled -coil system by calculating the ratio of the frequency split existing between the co and counter rotating mode of each system over their center frequen cy (or frequency at which the two coils of the inductively-coupled system resonated at individually) as the frequency split does depend on the coupling between the two coil of a system. 5 3 4 Samples A ll the samples were placed in 10 mm -diameter thin -wall glass tube s (Wilmad Labglass, Buena, NJ) to test the coil systems under unloaded conditions on the bench and in the magnet. The implantable coils were placed around this sample no more than a 1 c m away from the surface coil. Water was used for 1H detection, p e rfluorocarbon (PFC) sample s for 1 9F detection,
140 and phosphoric acid sample for 31P detection. If high concentration sample were initially used to test the functionality of the systems, more realistic samples were generated afterwards to define the sensitiv ity and limit of detection of the different coupled -coil systems for in vivo use under loaded conditions 5 3 4 1 Perfluorocarbons The PFC introduced in the samples of this study f or 19F detection was p erfluoro - crown 5 ether (PFCE, Exfluor Research Corp., Rou nd Rock, TX). It was selected for its single sharp 19F resonance signal from its 20 equivalent fluorine nuclei. It was prepared following a protocol adapted from Joseph et al. (169) Egg yolk lecithin (95 mg, Sigma -Aldrich Corp., St. Louis, MO ) was added to 600 mL of Tyrodes salt solution kept at 0 C (pH 7.4, Si gma -Aldrich Corp., St. Louis, MO) as an surfactant. While kept in an ice bath, this mixture was sonicated twice at 300 W for 15 s with a 1 min interval between sonications (Branson Ultrasonics, Danbury, CT ) to create an aqueous phase of lecithin. A volum e of 400 L of PFCE was then added to the mixture. Sonication was repeated seven more times under the conditions described above to disperse the PFCE in the mixture. This emulsion contained PFCE spheroid s bet ween 200 and 300 nm in diameter (Figure 53 A ). The emulsion was filtered through a 0.4 m membrane filter to make it sterile without destroying the preparation. It was shown not to affect the cells or their surrounding up to 10% (v/v) (170) Different amounts of emulsion (0.4%, 1%, 2.5 %, 5%, and 10%) were then incorporated to a alginate solution. Cell -free PFC a lginate beads were then generated as described by Simpson et al. (Figure 5 3B and (151) ). The alginate solution was a 2% (w/v) sodium alginate solu tion generated from alginate powder ( Novomatrix, Trondheim, Norway) and a solution of 0.85% sodium chloride
141 Aliquots of freshly made beads (~0.3 ml each) were transferred into the coil -construct assembly with an implantable coil resonating at the desired frequency for 19F detection us ing a syringe and 16 G needle (Figure 5 4). Figure 5 3. Photograph of PFCE emulsion. A) In solution; and B) in beads. (Photo by N.A. Volland). Figure 5 4 Implantable coil bioartificial pancreas macroconstruct asse mbly schematic and dimensions. A) Transverse view and B) coronal view of the assembly. The alginate beads are contained in the cylindrical cavity created by the PDMS -coated circular MR implant able coil and 300 m mesh screen
142 5 3 4 2 Phosphorus sample s In vivo p hysiological concentrations of phosphorus containing metabolites do not exceed 10 mM. Consequently, to assess the sensitivity of 31P detection, a sample of saline solution containing different concentration down to 500 M of sodium phosphate (Na2HPO4 2) a nd adenosinetriphosphate (ATP) were created, before cell -containing beads were tested in vitro To make the cell -containing beads for in vitro testing, m u -tet cells (48) were cultured and collected as described in S ection 4.3.5. Cell -containing a lginate beads (with a initial density of 3.5x107 cells/ml of alginate) were then generated and inserted in a coil construct assembly with an implantable coil resonating at the desired frequency for 31P detection as described in S ection 188.8.131.52 (Figure 5 4). The cell -containing assembly was placed in a 3 cm diameter flat -bottom Plexiglas tube closed with a vented cap The cells were f ed every 2 3 days and kept alive for the duration of the study. 5 3 5 NMR M easurements NMR measurements were performed on an 11.1 Tesla ( T ) 40 cm clear horizontal bore Magnex magnet equipped with a Bruker Avance c onsole Data acquisition was done using Bruker P aravision software as well as the imaging data processing. The spectroscopy data processing was done via a Bruker XwinNMR software. The signal to -noise ratio (SNR) for each measurement was then determined for each coil system with a custom image analysis software written in IDL (ITT Visual Information Solutions, Boulder, CO) when images could be acquired or the X w in NMR software when spectrum were acquired. When characterizing the single -frequency coil systems for 19F detection, 19F images were acquired using a spin -echo (SE) pulse sequence with sinc3 pulses, a repetition time (TR) of 1000 ms an echo time ( TE ) of 1 0 ms, 1 -mm slice thickness, 1 average 4 x 4 cm2 field of -view ( FOV ) and a 256x256 matrix on pure PFCE sample When characterizing the single -f requency coil
143 systems for 31P detection, 31P images were acquired using a spin -echo (SE) pulse sequence with sinc3 pulses, a repetition time (TR) of 1000 ms an echo time ( TE ) of 1 0 ms, 1 -mm slice thickness, 1 average 4 x 4 cm2 field -of -view ( FOV ) and a 64x 6 4 matrix on a phosphoric acid sample. PDMS signal suppression was not necessary for 1 9F and 31P detection since PDMS does not c ontain th ese nuclei. For the limit of detection studies, images were acquired (19F images) when possible using a spin echo (SE) pulse sequence with sinc3 pulses, a repetition time (TR) of 1000 ms an echo time (TE ) of 1 0 ms and 2 x 2 cm2 field -of -view ( FOV ). The slice thickness (1 3 mm) the number of signal averages (up to 10), and the matrix size (from 256x256 down to 32x32) were v aried to obtain images with acceptable SNR (> 2). Spectra were also acquired using one -pulse sequence with a hard pulse and a repetition time (TR) of 1000 ms The number of averages was varied up to 256 to obtain spectra with acceptable SNR (>2) within 10 minutes. The longitudinal relaxation times (T1) of PFCE were then determined from either 19F image or spectr um a cquisition To allow this determination, a saturation recovery spin echo imaging pulse sequence was used with sinc3 pulses with an echo time ( TE) of 10 ms, and variable repetition time (TR) of 5000, 2500, 2000, 1500, 1000, 750, 500, 200 ms to acquire the imaging data when 2.5% PFCE emulsion or more was used (171,172) Otherwise, a non -selective inversion recovery spectrum pulse sequence with hard p ulse s a nd the inversion time (TI) of 5000, 2500, 2000, 1500, 1000, 750, 500, 200 ms w as used to acquire the spectral data (171,172) The slice thickness (1 3 mm), the number of signal averages (up to 3) and the matrix size (from 128x128 down to 32x32) were varied for the i maging series to be acquired within 30 minutes. Only the number of averages was varied up to 256 for the spectra to be acquired within 30 minutes as well. Spectra of ATP were acquired using one hard -pulse sequence with a hard pulse, a repetition time (TR) of 2 000 ms
144 The number of averages (up to 1024) was varied to obtain spectra with acceptable SNR (> 2) within 30 minutes. 5 3 6 SNR M easurement s and S tatistical A nalysis The results of th e characterization s presented in this study w ere compared to each other and to the results obtained in the characterization of inductively -coupled implantable coil systems dedicated to 1H detection Th e se results g a ve a reference for the multiple -frequency system s that have been developed in parallel of the single -frequency c oil systems Th e comparison would assess the losses of sensitivity associated with multiple -frequency inductively-coupled implantable coil system s. Five different inductively coupled systems were tested: 1) one single -frequency system for 1H detection (470.75 MHz); 2) one single -frequency system for 19F detection (442.86 MHz); 3) one single -frequency system for 31P detection (190.5 MHz); 4) one double -frequency system for both 1H and 31P detection; and 5) one double -frequency system for both 1H and 19F det ection. The SNR of the different coil -systems were evaluated from the NMR images, when possible, by choosing a signal region-of interest (ROI) at the IC position and a noise ROI outside of the sample as described in S ection 3.3.8 for inductively coupled i mplantable coil system for 1H detection To assess the performances of spectroscopy methods, the SNR was evaluated from the NMR spectr a as described in S ection 4.3.8 for receive -only inductively -coupled implantable coil system for 1H detection. Multiple images or spectra were acquired for each coil system to assess the reproducibility of signal detection and the sensitivity of the systems. All the results are indicated as a mean standard deviation (relative standard deviations are indicated in parenthe ses in percentage).
145 5 4 Results 5 4 1 RF Coil System The simulations of the single -frequency inductivelycoupled implantable coil systems for 19F and 31P detection were similar to the ones of the coupled -coil system presented in S ection 3.4.1 F igure 5 5 displays t he simulated voltage standing wave ratio (VSW R) plots of the double -frequency 1H -31P inductively -coupled implantable coil system. The four resonan t frequencies, 2 modes per resonan t frequency, can be clear ly seen. Figure 5 5 Voltage Standing Wave Ratio ( VSWR) of the double frequency 1H -31P inductively coupled coil system at both frequencies A ) Z oom in around 190.5 MHz (31P resonant frequency at 11.1 T), B) entire frequency sweep; and C) zoom in around 470.75 MHz (1H resonant frequency at 11.1 T)
146 Th e near magnetic field at these resonan t frequencies was also simulated (Fig ure 5 6 ). The co and counter rotating mode s are present at both frequencies. The double -frequency 1H -31P circuit matching to 50 coupled coil system at once. Consequently, the magnetic fields simulated at any resonant frequency of the system were lower than the ones generated by single-frequency inductively coupl ed systems and only had qualitative value for the different modes generated by the overcoupling at the two resonant frequencies. Furthermore, the double -frequency 1H -31P circuit matching for the 31P resonances was worse than for the 1H ones, which lead to weaker magnetic fields at 31P resonances than at 1H resonances. Figure 5 6. Near magnetic field distribution along the axis of symmetry of the double -frequency 1H -31P inductively -coupled coil system at 190.5 MHz and 470.75 MHz A) Distribution at 185.45 MHz in the (+) mode, B) distribution at 465.75 MHz in the (+) mode, C) distribution at 195.8 MHz in the ( -) mode, and D) distribution at 475.45 MHz in the ( ) mode. The position of the surface coil wa s set at 0 and the implantable coil one at 1 cm.
147 Fi gure 5 7 displays the simulated VSWR plots of the double -frequency 1H -1 9F inductivelycoupled coil system T he four resonant frequencies and 2 modes per resonan t frequency can be clearly seen as well However, this VSWR graph looks different from the 1H -31P system one. The frequency split at both 1H and 19F frequencies is not symmetric around 442.86 and 470.75 MHz compared to any other inductively-coupled coil system built, single or double -frequency. The asymmetry seems more pronounced at the lower fre quency (19F) than at the higher one (1H). At the higher frequency, the split only looked shifted by 4 MHz, whereas at the lower frequency, the counter rotating ( ) mode frequency seemed to have shifted away from the original lower 442.86 MHz frequency com pared to the co rotating (+) mode frequency. The counter rotating mode of the lower frequency and the co rotating mode of the higher frequency seem to interact with each other. This double -frequency 1H -19F d for any of resonant frequencies of the system either. Consequently, the simulated magnetic fields (Figure 5 8 ) at any resonant frequency of the system only had qualitative values for the different modes generated by the overcoupling of the system at the two resonant frequencies. The co and counter rotating mode s at both frequencies are present in this system. However, the observation made on the VSWR graph can be seen here too. The counter -rotating mode of the lower frequency (442.86 MHz, Figure 5 8C ) is largely affected by the co -rotating mode of the higher frequency (470.75 MHz, Figure 5 8D) as its magnetic field magnitude is 10 times lower than the other ones.
148 Figure 5 7. Voltage Standing Wave Ratio ( VSWR) of the double frequency 1H -19F inducti vely coupled coil system at both frequencies A ) Z oom in around 442.86 MHz (1 9F resonant frequency at 11.1 T), B) entire frequency sweep; and C) zoom in around 470.75 MHz (1H resonant frequency at 11.1 T) Single -frequency inductively -coupled coil systems for 31P and 19F detection have been successfully developed. The implantable coil (Figure 5 9) and the surface coil (Figure 5 10) of a double -frequency 1H -31P inductively -coupled prototype system were also designed and built. These systems were all built in their (+) mode only, because previous inductively -coupled implantable coil systems have shown a higher sensitivity in this mode than their counterpart (see S ection 3.3.4).
149 Figure 5 8. Near magnetic field distribution along the axis of symmetry of t he double frequency 1H -19F inductively -coupled coil system at 442.86 MHz and 470.75 MHz A) Distribution at 436.4 MHz in the (+) mode, B) distribution at 459.7 MHz in the (+) mode, C) distribution at 443.75 MHz in the ( ) mode, and D) distribution at 488. 8 MHz in the ( ) mode. The position of the surface coil wa s set at 0 and the implantable coil one at 1 cm. Figure 5 9. Photograph of the double -frequency 1H -31P implantable coil. (Photo by N.A. Volland).
150 Even though the double -frequency inductively-coupled coil system could be matched to 50 matching capacitors originally placed in the circuit, it was only for one frequency at a time. It was found that the matching at 31P frequency required high matching capacit ors values, while the matching at 1H frequency required smaller capacitor values. To compensate for this matching issue, a matching network was quickly implemented on the double -frequency surface coil (107) A ta nk circuit was used as a matching network (Figure 5 10). The initial matching capacitors were replaced by capacitors with smaller capacitance values ( Voltronics Corporation, Denville, NJ, 0.5 8 pF ) to allow the matching network to match the system to 50 at the higher frequency since the tank capacitor dominated the matching of the high frequency T he inductors were small solenoids made of 28 AWG copper wires, hand wound around a 16 AWG copper wire and placed over the small matching capacitors to allow the matching network to match the system to 50 at the lower frequency since the tank inductor dominated the matching of the system Figure 5 10. Photograph of the double -frequency 1H -31P surface coil. (Photo by N.A. Volland). No 1H -19F inductivel y -coupled implantable coil system was built. The fact that the ( ) mode of the lower frequency and the (+) mode of the higher frequency coupled to each other because of their closeness and their shift away from each other made it harder to actually tune t he system at the desired high frequency. Furthermore, the coupling capacitors (CTui and CK sV o n
151 Figure 5 2) were very small (< 1pF), which anticipated problems for the implantable coil to be properly tuned when coated and loaded. In order to build and tes t a 1H -19F system a better design will have to be investigated in the future Bench testing was performed and resonan t frequency and Q factor of the different coils built and tested are summarized in Table 5 1. The inductively-coupled systems for 31P det ection in the circuit at 190.5 MHz. To decrease the resistive loss in the circuit as much as possible, the wire thickness needed to be more than 5 times the s kin depth. This involved the replacement of the 35m thick copper tape of the surface coils by a round 20 AWG wire (0 812 mm diameter). The Q of the surface coils of these coupled -coil systems improved by at least 20% after this change and they are the only ones listed on Table 5 1 for these systems. The surface coil of the single -frequency inductively -coupled implantable coil system s has a Q of 72.2 4.95 ( 6.86%) when resonating at 493.2 MH z for 1H detection (n = 4) a Q of 63.3 when resonating at 464.27 MH z for 19F detection (n = 1) and a Q of 54.6 when resonating at 199.6 MH z for 31P detection (n = 1). The surface coil of the double -frequency inductively coupled implantable coil system has a Q of 50.81 when resonating at 475.1 MH z for 1H detection (n = 1) and a Q of 34.30 when resonating at 19 5.5 MH z for 31P detection (n = 1). The implantable loop -gap resonator has an average Q of 203.90 6.20 ( 3.04%) when resonating at 491.81 3.38 ( 0.6 9 %) MHz for 1H detection (n = 7) an average Q of 171.86 13.78 ( 8.01%) when resonating at 467.22 2.49 ( 0.53%) MHz for 19F detection (n = 3) and an average Q of 122.9 5.18 ( 4.22%) when resonating at 202.6 0.14 ( 0.53%) MHz for 31P detection (n = 3) The implantable loop gap resonator of the dou ble -frequency inductively coupled implantable coil system has an average Q of 20.26 4.34 ( 21.44%) when resonating
152 at 49 1.27 1.75 ( 0.36%) MHz for 1H detection, and an average Q of 47.93 13.40 ( 2 7.95%) when resonating at 202.35 2.05 ( 1.01%) M Hz for 31P detection (n = 2). The single frequency inductivelycoupled implantable coil system has a n average Q of 96.21 3.70 ( 3.85%) when resonating at 470.75 MHz for 1H detection (n = 7) a n average Q of 85.56 3.00 ( 3.51%) when resonating at 4 42.86 MHz for 19F detection (n = 3) and a n average Q of 73.23 4.60 ( 6.28%) when resonating at 190.5 MHz for 31P detection (n = 3). The double -frequency inductively-coupled implantable coil system has a n average Q of 31.05 0.49 ( 1.59%) when resonati ng at 470.75 MHz for 1H detection and a n average Q of 36.25 0.35 ( 0.95%) when resonating at 190.5 MHz for 31P detection (n = 2) The single -frequency inductively -coupled coil systems have an average coupling factor (k) of 0.082 0.0035 ( 4.22%) deter mined from the frequency split around the desired resonant frequency obtained on the simulated voltage standing wave ratio (VSWR) graph from the systems for the detection of 1H, 19F, and 31P. On the bench this average k became 0.086 0.007 ( 7.63%). Th e double -frequency 1H -31P inductively -coupled coil system, however, has a different coupling factor (k) for both frequencies when determined from the frequency split around the desired resonan t frequency obtained on the simulated voltage standing wave rati o (VSWR) graph : 0.055 at 190.5 MHz (31P ) and 0.021 at 470.75 MHz (1H) On the bench this average k became 0.074 0.0023 ( 3.19%) and 0.049 0.0001 ( 0.28%) respectively. The double -frequency 1H -19F inductively -coupled coil system also has a different coupling factor (k) for both frequencies when determined from the frequency split around the desired resonant frequency obtained on the simulated voltage standing wave ratio (VSWR) graph : 0.017 at 442.8 MHz (19F) and 0.062 at 470.75 MHz (1H) This last f actor is biased by the coupling occurring between the ( ) mode of the lower frequency and the (+) mode of the higher frequency.
153 Table 5 1 Implantable coil, surface coil, and inductively -coupled implantable coil system characteristics Resonan t frequency (f) and quality factor (Q) for : SC IC and coupled coil systems for 1H detection (SCH ; ICH ; SH ), 19F detection (SCF ; ICF ; SF ), 31P detection (SCP ; ICP ; SP), and 1H -31P simultaneous detection (SCHP ; ICHP; SHP). Coil #* # of coils tested f (MHz) Q SCH 4 4 93.2 72.2 4.95 ( 6.86%) SCF 1 464.27 65.25 2.76 ( 4.23%) SCP 1 199.6 54.6 SCHP_H 1 475.1 50.81 SCHP_P + 1 195.5 34.30 ICH 7 491.81 3.38 ( 0.6 9 %) 203.90 6.20 ( 3.04% ) ICF 3 468.39 1.55 ( 0.33%) 182.54 3.55 ( 1.95% ) ICP 3 202.60 0.1 4 ( 0.07 %) 122.90 5.18 ( 4.22% ) ICHP_H 2 491.27 1.75 ( 0.36%) 20.26 4.34 ( 21.44%) ICHP_P 2 202.35 2.05 ( 1.01%) 47.93 13.40 ( 27.95%) SH 7 470.75 96.21 3.70 ( 3.85%) SF 3 442.86 80.4 3.11 ( 3.87%) SP 3 190.5 73.23 4.60 ( 6.28% ) SHP_H 2 470.75 31.05 0.49 ( 1.59%) SHP_P + 2 190.5 36.25 0.35 ( 0.97%) *The SC and IC were tested individually unloaded. The inductivelycoupled implantable coil systems were tested using the 1cm diameter tube water sample. + The double frequenc y 1H -31P surface coil and inductively coupled coil system could not be matched to 50 (or 20dB). Consequently, the Q indicated above was determined when the coils had a match around 10 dB and not comparable to the other coils characteristics (144) T he double -frequency inductively-coupled implantable coil system did not pass the test since the system could not be match to 50 for both frequencies at once when using the
154 matching network. The matching network will have to be optimized before this system can be tested further. Using variable inductors instead of hand -woun d inductors would be the first improvement to be investigated. 5 4 2 Implantable Coil Coating The single -frequency 19F and 31P implantable coils were coated with PDMS for in vitro testing as the double -frequency systems did not pass the bench test. Similar to t he 1H implantable coil coating, a 1 -mm PDMS layer and a 2 -step process were used as described in S ection 184.108.40.206. The frequency shift observed for the single frequency 1H implantable coils was also observed for the 19F coils. A resonant frequency shift o f 6.46 0.82 ( 12.75%) was detected for the coated implantable coils compared to the non-coated similar coil (n = 5). Three coils were found with a defective coating and were removed from the experiments. The quality factor of these coated coils was b etter than the one of the uncoated coils (22% higher on average) as the frequency of the coated coil shifted down compared to the uncoated coil and its bandwidth decrease slightly As mentioned in S ection 3.5 this indicates that the coating introduces more capacit ance in the circuit than resistance As loading further changes the coil characteristics, the se 19F coated implantable coil s were embedded in the mouse abdomen -like phantom Th e loading caused an additional resonan t frequency shift from the PDM S coating one of 28.19 2.48 ( 8.78%) MHz. This shift in addition to the coating shift was taken into account to build the 19F implantable loop -gap resonator coils which were used for in vitro and in vivo studies. The quality factor of these loaded coated coils decreased by a factor 9.5 compared to the coated coil Q determined in air (17.85 1.92 ( 10.77%)). The implantable coils to be used loaded were created with a resonan t frequency at least 3 5 MHz higher than their predecessors ( 506.12 5.20 ( 1.03% ); n = 7 ). This made the inductively -coupled 19F coil systems resonate at the
155 desired frequency of 4 42.86 MHz after coating and loading with media culture and ready to be tested in the magnet for in vitro studies. The frequency shift caused by the coating and observed for the single -frequency 31P implantable coils was very s mall [ 1.71 1.28 ( 74.94%) ] compared to the two other higher frequency implantable coils (n = 6). The quality factor of these coated coils was slightly lower than the one of the uncoated coils (9% lower on average). As loading further changes the coil characteristics, these 31P coated implantable coils were embedded in the mouse abdomenlike phantom. This loading caused an additional resonant frequency shift from the PDMS c oating one of 2 .58 0.3 ( 11.76%) MHz. This shift was small again The quality factor of these loaded coated coils decreased by a factor 2.75 compared to the coated coil Q determined in air [42.85 3.01 ( 7.03%) ]. Combining the coating and loading effects, the implantable total frequency shift was below 5 MHz. This allowed the coils to be used for in vitro and in vivo studies without even altering them and made the inductively -coupled 31P implantable coil systems resonate at the desired frequency of 190.5 MHz after coating and loading and ready to be tested in the magnet for in vitro studies. 5 4 3 Sample S tudies The coupled -coil systems used in th e magnet study include d : 1) one single -frequency coupled coil system for 19F detection (442.86 MHz); 2) one single -frequency coupled -coil system for 31P detection (190.5 MHz). The single -frequency coupled coil system for 1H detection (470.75 MHz) developed in Chapter 3 was used as a reference. Images were successfully acquired using all systems on their respec tive high concentration samples under unloaded conditions. 19F images and spectra were also achieved from the beads with PFCE emulsion, and 31P spectra from cells in alginate beads w ere attempted The multiple -frequency
156 1H -31P inductively -coupled implanta ble coil system was not tested in the magnet as the matching of the system to 50 was n ot achieved in the bench. 5 4 3 1 Perfluorocarbon d etection For the different concentrations of PFCE emulsion ( 0.4%, 1%, 2.5 %, 5%, and 10% (v/v) /mL of alginate) used, images and spectra were acquired from the cell -free bead samples. 10% PFCE emulsion was given the most signal for both images and spectra. However, 2.5% PFCE emulsion was thought more appropriate to use for imaging as it is a relatively low PFCE content and it still allow an SNR of 5.15 0.29 ( 5.56%) within 4 minutes (Figure 511). This result would potentially allow T1 mapping imaging throughout the construct in the future. It is worthwhile noting that non -spatially selective sequences were used to acquire these 19F images and spectra and no signal coming from outside the construct was detected (Figure 5 11) 19F spectra were also acquired with acceptable SNR (> 2) with all concentration of PFCE within a maximum of 8 minutes. When using the 0.4% PFCE emuls ion, the spectra were acquired in 4.25 minutes with the SNR over 100 when using 256 averages. The longitudinal relaxation time (T1) of PFCE could not be determined from a series of images acquired with 2.5% PFCE emulsion within 30 minutes, the SNR may hav e been too low to allow the program to properly determine the value, since the results obtained for T1 were unrealistic. However, PFCE T1 w as estimated at 940 ( 8.7x103) ms from spectrum data of a 0.4% PFCE emulsion entrapped in beads within 30 minutes (Figure 5 12)
157 Figure 5 11. 19F images of a pancreatic construct surrounded by an implantable coil and containing BaA beads with 2.5% PFCE. A) Transversal view; B) sagittal view; and C) coronal view. An SE pulse sequence was used to acquire these ima ges: TR = 1000 ms, TE = 10 ms, 2 -mm slice thickness, 3 average s FOV 2x2 cm2 and matrix size 64x64.
158 Figure 5 12. 19F spectrum series from BaA beads containing 0.4 % PFCE within a pancreatic construct surrounded by an implantable coil. A series of inv ersion recovery pulse sequence s w ere used to acquire these spectra in 30 mins (TR = 5 0 13 ms, TI= 5000, 2500, 1500, 1000, 750, 500, 250, 10 ms, 256 average s ) using an inductively -coupled implantable coil system for 19F detection under loaded conditions 5 4 3 2 Pho sphorus d etection S pectra were successfully acquired from the solution samples with different concentrations of Na2HPO4 2 and ATP (100mM, 10mM, 5mM, 1mM and 0.5 mM ). Figure 5 13 displays a 31P spectrum obtained from a 0.5 mM of sodium phosphate and ATP s olution The spectrum was acquired in 35 minutes using a one hard -pulse pulse sequence with TR = 2s and 1024 averages and an inductively -coupled implantable coil system for 31P detection under unloaded conditions. The SNR displayed on this spectrum was de termined at 3.2, which is encouraging even though under unloaded conditions because physiological ATP level should be between 1 and 10 mM. S pectra acquired from TC -tet cells encapsulated in BaA beads did not display more than the signal from the inorganic phosphate (Pi) contained in the culture media. This proved that the inductively-coupled implantable coil system for 31P detection worked. However, the viability of the entrapped cells was questionable. The culture m edia was used during the magnet testi ng as TI increase
159 none of its compounds was known to interfere with ATP detection. Howeve, i t was later found that the pH of the media changed rather rapidly when in a low CO2 concentration environment and this could have killed the cells Cell density may have also been an issue in this early study on ATP detection. The number of cells in the sample may have also been too low to be detected. Figure 5 1 3 31P spectrum of a saline solution containing 500 M of ATP and PO4 2. The spectrum, displaying the peaks of the three different phosphate groups of ATP and inorganic phosphate (Pi), was acquired using a n inductively -coupled implantable coil system at 11.1 T within 35 minutes with an SNR of 3.2 5 5 Discussion Monitoring the viability and the bioenergetic s status of cells as well as the oxygen concentration within a tissue -engineered construct, such as a bioartificial pancreas, using NMR techniques is a powerful tool to directly reflect the well being and functioning of the construct. In order to achieve such a comp lete monitoring, the use of different single -frequency inductively-coupled, implant able coil system s for 1H, 19F, and 31P detection to perform NMR
160 measurements was explored The development of double -frequency inductively -coupled implantable coil systems for simultaneous detection of 1H and 19F, and 1H and 31P detection was investigated in parallel to the construction of single -frequency coupled -coil systems. The single frequency systems would serv e as standards to assess the efficacy of the double -freque ncy sy stem s Th e double -frequency coupled coil sy stem s ha ve the following advantages over single frequency systems for the study of tissue -engineered constructs: 1) simultaneous NMR measurement s from 1H, 19F, and 31P nuclei; and 2) direct correlation of t he NMR results obtained from the detection of the different nuclei under investigation. The well being and appropriate functioning of a bioartificial construct in its environment and the time line prediction of its potential failure are key in monitoring s uch an implanted device Obtaining NMR measurements for different nuclei helps reach these goals. The use of single -frequency inductively -coupled coil systems has shown a high sensitivity for 1H detection (Chapter 3) The use of these systems for 19F an d 31P detection here also showed to be of high sensitivity in phantoms. A concentration of 2.5 % PFCE emulsion was detected in vitro with both 19F MRI and MRS with an SNR above 5 for imaging and above 100 for spectr oscopy. Lower than physiological levels of ATP were also detected with a SNR over 3 under unloaded conditions. I f these levels of detection of fluorine and phosphorus (19F and 31P) are encouraging, the use of multiple -frequency inductivelycoupled implantable coil systems is more desirable tha n the use of single -frequency inductively -coupled system as only one coil can be implanted within a construct in vivo and the correlation of the results would be made easier. However, their construction requires several considerations as a large number of parameters intervene when the coils have two resonant frequencies each. The main parameter was the ratio of the tank circuit inductance to the coil/sample inductance (main circuit
161 inductance) To engender fewer resistive losses in the main circuit at the lower frequency (less sensitive nucleus) compared to the higher frequency (107) th is ratio was kept at 5 in the case of the 31P -1H inductively -coupled implantable coil system since the lower frequency was much le ss sensitive than the higher one. This also kept the circuit sensitivity at lower frequency close to 90%. However, overall resistive losses at each frequency for a multiple-frequency tank system are always higher than the losses at the same frequency for a single -frequency system (118) In the development of 19F -1H inductively-coupled implantable coil system; the tank induct ance to main circuit induct ance ratio was kept close to 1 since the frequencies had almost the same h igh sensitivity (167) and the resistive losses were going to be comparable (about 20% for both) Even though the multiple -frequency inductively-coupled implantable coil systems were not tested in the magnet, t he ir higher electrical losses than the single frequency system may have been observable through their Q and coupling factor (k) determined at the system different frequencies. The coupling factor is defined as a geometry factor (Equation 5 1) which was easily verified with the different single -frequency inductively-coupled co il systems developed at different frequencies in this study. However, it does not appear to be the same for multiple frequency inductivelycoupled coil systems as k of both frequencies of the system is smaller than k of the single -frequency inductively-co upled coil system when the geometry of all the systems is identical. Furthermore, the higher frequency had a smaller coupling factor than its lower frequency counterpart. It is possible that the simple relationship between the mutual inductance and the c oupling factor of the system does not hold for double -frequency inductively-coupled coil system because the system inductances and mutual inductance has a complex dependence on more than one inductor and different at both frequencies. It is also worthwhil e noted that the Q
162 of the double -frequency inductively-coupled implantable coil system is lower than that of its single -frequency counterparts at both frequencies. The construction of multiple -frequency implantable coils was rendered difficult because of t he fixed capacitor used. Even though the tuning capacitor mainly a ffects the tuning of the lower frequency and the tank capacitor the tuning of the higher frequency, the discrete capacitor values available prevented tuning to the exact frequency required. Multiple capacitors configured in parallel could have facilitated obtaining the right combination, but size restriction will have to be considered before doing this. Furthermore, matching the circuit to 50 for both frequencies simultaneously was rende red difficult by the use of fixed inductors. Variable inductors may be preferable to allow fine tuning with different loading and need to be considered for further work on the multiple -frequency inductively -coupled coil system. The study of the coating and loading effects on single -frequency implantable coils showed that coating and loading have larger effects at higher frequencies than at lower ones. Even though working at higher field means better sensitivity, this could be an advantage for multiple freq uency inductivelycoupled systems as 31P (less sensitive nuclei) will not be affected as drastically as 19F or 1H at 11.1 T. In the sample studies, the SNR and acquisition time of 19F images obtained using a 2 .5 % PFCE emulsion entrapped in beads confined in the construct indicated the possibility of generating T1 mapping images within 30 mins. Such a measurement would carry information on the oxygenation throughout the constructs in vivo at different time point during a study period. However, when attemp ted, the images acquired in the imaging series to determine T1 had a too low SNR to allow an accurate determination of T1. No image series could be acquired in a reasonable amount of time when using a 0.4% PFCE emulsion. Nonetheless, a series of spectra
163 was acquired from a construct containing beads with 0.4% PFCE emulsion in vitro and T1 of PFCE was determined accurately (940 ms). This value represents the average longitudinal relaxation time of PFCE throughout the construct under normoxic conditions. It is in agreement with the estimation done by Gross et al. at a similar magnetic field strength (11.7T) with a similar sample (0.4% PFCE). Furthermore, t he 19F detection sensitivity was kept constant over at least 2 weeks (95) This demonstrates the possibility to conduct stu dies on the evolution of the oxygen level within the pancreatic constructs over time in vitro and in vivo using a single or double -frequency inductively -coupled coil system and MRS localized or not Either the concentration of PFCE in the beads or the se nsitivity of the method will ha ve to be increased, or localized spectroscopy will have to be implemented to be able to perform T1 mapping and study the oxygenation in different region s of the construct over time. The SNR displayed on the ATP spectrum ( 3 .2) acquired from a 0.5 mM ATP solution shows adequate sensitivity (SNR > 2) to perform studies on cell bioenergetics in vitro and in vivo over time using a single -frequency implantable coil system. However, no ATP signal could be detected from cells entrapp ed in beads confined in a coil -construct assembly. The cells viability was questionable during the NMR experiments and will ha ve to be checked more closely in the future since ATP washed out relatively rapidly after cell death. The cell density within t he con struct may also be of concern and will need to be increased in future experiments, at least initially to optimize the method and determine the limit of sensitivity of the method. O ther groups have successfully performed 31P MRS at the same field str ength (11.1 T) using the same pulse sequence and parameters in combination with a surface coil of similar size (173) but less sensitive than an inductively -coupled implantable coil system ( Chapter 3). Coupled -coil system design adaptation will have to be made prior to a double -frequency inductively -coupled
164 implantable coil system development since its design will only preserve 90% of 31P detection sensitivity from a single frequency system (118) 5 6 Summary Inductively -coupled i mplantable coil system for 1H, 19F, 31P, simultaneous 1H and 31P, and simultaneous 1H and 19F detection have been developed and their performance analyzed. 1H, 19F, 31P, and 1H -31P systems were successfully built and characterized on the bench. NMR images and spectra were also obtained at 11.1 T with the three single -freq uency coupled -coil systems demonstrating the capability of inductively-coupled implant able coil systems to monitor bioartificial pancreas constructs using different nuclei detection at high field. 19F NMR imaging was successfully performed on a 2 .5 % PFCE e mulsion sample when a single -frequency inductively -coupled coil system for 19F detection was used. These images also show ed that no localization was necessary when performing 19F spectroscopy since fluorine was confined to the bioartificial construct 19F spectra were also easily acquired with even lower PFCE emulsion concentration (0.4%) under loaded conditions allowing the determination of PFCE T1. This attest s the feasibility of monitoring the evolution of the oxygen level within the construct via spe ctrum series acquisitions. Similarly, ATP was found to be detectable at cellular concentration via NMR spectroscopy using a single -frequency inductively-coupled implantable coil system for 31P detection. These results suggest the performance standards fo r the development of multiple -frequency inductively -cou pled implantable coil systems. The feasibility of d ouble -frequency inductively-coupled implantable coil system s for the detection of 1H, 31P, and 19F w ere also explored, as the se systems would allow th e simultaneous detection of key nuclei of different sensitivity yet critical to quantify implanted bioartificial organ function. The development of such systems thus far shows that they are feasible using tank circuit theory even though these systems a re complex and requires lots of con s i de rations.
165 They also show ed their potential over single -frequency coupled -coil systems, as they can detect more than one nucleus at a time and allow direct correlation of the results from these multiple detections. T h e pursuance of their development is encouraged by the possibility of develop ing a triple frequency inductively -coupled implantable coil system in the future E fforts c an now focus on further development s of the double -frequency inductively coupled implanta ble coil system s to allow the detection of more potential early marker of pancreatic construct failure before hyperglycemia could be detected. However, having a functioning construct is also a major consideration for future stud ies on the noninvasive mon itoring and complete characterization of a bioartificial construct Consequently, g enerating cell -containing bioartificial construct and keeping the insulinproducing cells alive within a bioartificial construct (including an implantable coil) in vitro an d in vivo was studied next.
166 CHAPTER 6 CHARACTERIZATION OF A BIOARTIFICIAL PANC REAS AS PART OF A N IMPLANTABLE COIL -CONSTRUCT ASSEMBLY 6 1 Introduction Bioartificial pancreatic substitutes, generated by either microencapsulation or macroencapsulation, have already shown potential as alternative to current treatments of type 1 diabetes. Non invasive, high -sensitivity monitoring techniques, like NMR imaging and spectroscopy using inductively -coupled implantable coil systems are well -suited and preferable to directly moni tor these pancreatic construct s their performances, and life span in a given environment. However, the inclusion of the implantable RF coil within the bioartificial construct places some restrictions on the construct design. In this study, two flat -disk macroconstruct s were studied: (1) one made with alginate foam, and (2) one made with alginate beads. Their design and materials, their construction, and inclusion in the middle of an implantable RF coil were investigated. Their effects on the entrapped TC tet cells viability and function and their ability to reverse a streptozotocin -induced diabetic state in mice were evaluated. The data show that a coil -construct assembly allowed insulin-producing cells to function and stay viable for extended period of time. The implantable coil bioartificial construct assembly implantation in a mouse peritoneal cavity also demonstrate s the return of normoglycemia in diabetic mice while allowing for enhanced non -invasive monitoring of implant ed constructs using NMR methods. Assembly pre -treatment and coating are also discussed to decrease construct fibrotic overgrowth. 6 2 Background Tissue -engineered pancreatic constructs emerged over twenty -five years ago. Bioartificial pancreatic substitutes, also called bioartifici al pancreas (BAP), have already shown potential as an alternative to current treatments of type 1 diabetes. These constructs are comprised of individual insulin -producing cells or islet like clusters of insulin -producing cells within a
167 biocompatible, semi permeable membrane. The membrane provides mechanical protection and partial immunoisolation. Optimally, it should be permeable to low molecular weight nutrients and metabolites, but impermeable to higher molecular weight molecules, such as immunoglobulin and albumin. Current techniques to generate these constructs are microencapsulation (29,34) and macroencapsulation (5,6) Microencapsulation creates microspheres or beads with diameter from a fraction of a millimeter to a couple of millimeters (33) Their small size makes t he capsules implantable almost anywhere in the body with a simple needle, the peritoneal cavity being the main and easiest implantation site (33,35,36) Alginate has often been used as the semipermeable membrane for microencapsu lation (33,37, 55) Its structural integrity, ease of handling, and diffusion properties allow the entrapped cells to live and function as normally as possible. One drawback of these microconstructs is their dispersion throughout the body making in vivo close monitori ng and retrieval of this type of construct nearly impossible. Macroencapsulation generates large semipermeable capsules in which a large number of cells can be suspended (38 41) The macroconstruct semipermeable matrix can be made with different polymers (38,41) or nuclepore membranes can be mounted on a silicone ring (39) They can take the shape of a tube or a planar sheet. Their size renders them confined to their site of implantation and easy to locate, monitor directly, and retrieve. Furthermore, t he use of higher magnetic field strength along with inductively-coupled implantable coil systems showed a great improve ment in the sensitivity of NMR imaging and spectroscopy to noninvasively monitor a bioartificial pancreas (see Chapter 3). However, the inclusion of the implantable RF coil within the BAP places some restrictions on the construct design. Per coil design, a flat -disk BAP represents th e optimal design for this approach.
168 This chapter investigates two type of flat -disk bioartificial construct design amenable for inclusion in the implantable RF coil: (1) one implantable coil alginate foam assembly and (2) one implantable coil alginate bea d assembly The preliminary in vitro and in vivo studies presented in this chapter outline the design, construction, and characterization of a BAP and its inclusion in the middle of an implantable RF coil. It reports the impact of the materials used to b uild the construct on the entrapped insulin-secreting cells (viability essentially) and on the coil (stability), and discusses technical aspects important in developing a construct with appropriate structural integrity and adequate cell function while bein g monitored directly via NMR. 6 3 Materials and M ethods 6 3 1 Cell C ulture tet cells (48) were obtained from Shimon Efrat s lab oratory (Albert Einstein College of Medicine, Bronx, NY ). They were cultured in T -flasks in monolayers an d fed every 2 -to 3 days with high glucose Dulbeccos Modified Eagles Medium (DMEM Invtrogen Corp., Carlsbad, CA ), supplemented with 10% fetal bovine serum (FBS, Lonza Williamsport, PA) 1% penicillin -streptomycin (Cellgro, Manassas, VA ) and 1% L -gluta mine (Cellgro, Manassas, VA ) until they reached confluence. Cultures were maintained at 37C under humidified and gas -regulated (5% CO2 / 95% air) conditions in an incubator. The cells were collected from the flasks for construct fabrication u sing 0. 01% Tryps in (Sigma, St. Louis, MO) Cell passages from 31 55 were used for these studies. 6 3 2 Implantable Coil -P ancreatic C onstruct A ssembly Two types of pancreatic constructs were generated from the ring formed by the PDMS coated coil presented in S ection 3.4.3: 1) including alginate sponge or foam disks (149,150) ; and 2) including alginate beads (33,37,55) The diam eter of the construct did not exceed 1 cm in
169 order to fit within the middle of the coated implantable coil and its thickness did not exce ed 500 m to minimize the diffusion limitations. 6 3 2 1 Implantable c oila lginate foam a ssembly Alginate foam was purchased (N ovomatrix, Trondheim, Norway) Three types of foam were considered: 25%, 50%, and 75% calcium -cation saturated. Circular, 1 -cm diameter pieces were punched out of this foam and placed in the middle of the PDMS -coated coil (Figure 6 1). The ability of th e different foams to swell or shrink was tested first by soaking them in PBS. The piece of foam with the minimal deformation capability in solution was selected as this can largely influence the viability of the cells and stability of the construct during the course of an in vitro and in vivo study. This piece of foam was then fixed in the middle of a PDMS -coated coil with nonabsorbable surgical sutures prior to in vitro testing. Figure 6 1. Photograph of an implantable coil alginate foam construct assembly prior to being soaked in PBS. (Photo by N.A. Volland). 6 3 2 2 Implantable c oila lgin ate b eads a ssembly To keep the small -diameter beads together in the middle of the implantable coil, a biocompatible polyetheretherketone (PEEK) 300 m mesh (Small Parts Inc., Miami Lakes, FL) was added on the top and bottom of the ring formed by the PDMS -coated coil (Figure 6 2). The cavity created, similar to the construct developed by Hirotani et al. (39,56) was 4mm high and 10mm in diameter, or a volume of 0.31 mm3
170 very fine lay er of PDMS. A small portion of one of the meshes was left unattached to allow the bead insertion after their creation via a 16G needle. For in vitro studies only, three 2 -mm diameter stainless steel spheres were attached with the mesh to one side of the meshed PDMS ring to create a stand and help with the homogeneous diffusion of nutrients and oxygen through the construct. Figure 6 2. Implantable coil -bioartificial pancreatic construct assembly diagram and dimensions. A) Transverse view and B) corona l view of the assembly. The alginate beads are contained in the cylindrical cavity created by the PDMS -coated implantable coil and a 300m mesh screen. All the implantable coil construct assemblies were first acid washed and soaked in a 0.1 M solution of hydrochloric acid for 30 minutes to break down the proteins left on the coil during manufacturing then soaked with a 75% ethanol solution for one hour to neutralize the broken down proteins These assemblies were then rinsed with distilled water to wash out the proteins and autoclaved for sterilization prior to in vitro or in vivo use. 6 3 3 Cell Entrapment 6 3 3 1 Alginate foam The coil -construct assemblies were first soaked into culture media for 2 hours to get the foam wet prior to in vitro studies. PBS was used a s a wetting agent prior to in vivo studies. Freshly trypsinized c ells (3.5 x107 cells/ml alginate) were then seeded at the top and /or bottom of the foam via a micro pipette.
171 6 3 3 2 Algin a te beads A low viscosity (325 mPa) high (62%) mannuronic / low ( 38%) g uluronic (LVM) alginate (NovaMatrix, Oslo, Norway) was used to generate the beads (174) The a lginate solutions were prepared by dissolving LVM alginate powder into physiological saline (0.85% NaCl) at a 2% w/v concentration overnight Both cell -free and cell -containing beads were prepared. F reshly trypsinized cells were mixed with the 2% alginate solution right before making the cell containing beads. Beads were generated by running the alginate solution (with or without cells) through a syringe at a constant flow rate and dropping it in a cationic cross linking solution with an electrostatic bead generator (Nisco, Zurich, Switzerland) to set the bead diameter. The bead diameter was kept between 400 and 600 m as p hysiology indicates that confluent cells should be within 200 250 m of an oxygen and nutrient access for sufficient oxygenation and adequate nutrient supply essential to their well -being The number of cells in the solution at the time of encapsulation was set at 3.5 x107 cells/ml of alginate Calcium cations from a 1 00 mM CaCl2 solution and barium cations from a 55 mM BaCl2 solution were indifferently used as crosslinkers Calcium -crosslinked alginate beads were then coated with poly -L lysine (PLL; Sigma, St. Louis, MO ) using protocols developed by Lim and Sun (34) and implemented regularly by our laboratory (151,175) to increase the bead stability and give them an accepta ble porosity. These beads were then dipped in a 0.2% w/v alginate solution (using the same alginate composition as the core alginate) to decrease the potential immune response to PLL and form the alginate/poly L -lysine/alginate (APA) beads. As barium is a stronger ionic crosslinker, no further processing was needed to produce b arium -crosslinked alginate (BaA) beads (176) The beads were only washed up to 3 times in BaCl2 solution of decreasing concentration to increase the saturation of alginate with Ba2+ and enhanc e the of the alginate network.
172 Both types of beads were rinsed three times in cult ure media after manufacturing to make sure all potential harmful compounds used in the process were removed from the solution. Different a liquots (up to 0.3 mL) of freshly made beads (APA or BaA ) were transferred in the middle of the meshed PDMS -coated co il ring constructs directly after manufacturing 6 3 4 C ulture P ropagation For in vitro studies, t h r e e similar cell -containing implantable coil construct assemblies w ere placed into one high -walled 60 mm Petri dish, fed with complete DMEM every 2 3 days and pl aced on a platform rocker (Stovall Life Sciences, Greensboro, NC) in the incubator The sensitivity of temporal glucose changes was increased by this practice. T he error in cell number estimation was also decreased To reduce immune response to implanted coil -construct assemblies, construct made for in vivo work were never rinsed with culture media at any time during the preparation, but with sterile phosphate buffer saline (PBS, Invitrogen, Carlsbad, CA), supplemented with 20 mM glucose antibiotics (100 U/ml penicillin, 100 ng/ml streptomycin, and 100 nM dexamethas one ), and buffering agent (20 mM HEPES). T he constructs were kept in this solution until implantation in a mouse peritoneal cavity Implantation occurred less than 2 hours after the construct preparation ended 6 3 5 Glucose C onsumption Cell g lucose consumption rates (GCR) were determined for every in vitro study at different point in time to determine cell growth. S amples of the culture media were collected from each assembly -containing Petri dish immediately after every feeding (T0) and 24 h ours later (T1) during the in vitro studies Sample g lucose levels were measured on a Vitros DT60II bioanalyzer (Ortho Clinical Diagnostics, Rochester, NY). Glucose consumption rates (GCR)
173 were then determine d by calculating the difference between the sample glucose level at T0 and T1. The results were normalized to 105 cells at the time of encapsulation. 6 3 6 Histology Histology samples of alginate bead s were taken at the start and conclusion of the studies both in vitro and in vivo T hey were fixed in 3% glutaraldehyde for at least 10 mins upon retrieval. They were then rinsed in successive ethanol washes (25%, 50%, 75% EtOH) over the next 3 hours, embedded in paraffin, sliced, and stained with hematoxylin/eosi n (H/E). Beads retrieve d from in vitro studies were first rinsed in their respective cationic solution (CaCl2 or BaCl2) before being fixed to increase the strength of the alginate bead network before histology processing. Digital photomicrographs were ta ken using a 10x objective (Nikon, Japan). 6 3 7 Streptozo to cin Injection a nd Blood G lucose M onitoring Diabetes was induced in normal female C3H/HeN mice through a series of streptozotocin (STZ; SIGMA, St. Louis, MO) injections of 100 mg/kg intraperitoneally (IP) over 5 days. The blood glucose levels in the mice were checked daily taking blood samples from the tail vein. The glucose levels were measured using a OneTouch Ultra2 (LifeScan, Milpitas, CA) blood glucose monitoring system. The mice were considered di abetic once their blood glucose levels exceeded 300 mg/dL of blood for three consecutive days. 6 3 8 Animal Preparation and Handling E xperiments with adult female C3 H/HeN mice, weighing 20 30 g, we re conducted according to a protocol approved by the University of Florida Institutional Animal Care and Use Committee. NMR measurements and s urgeries were completed under general anesthesia. Anesthesia was induced by inhalation of 2% isoflurane in oxygen and maintained for the duration of the surgery. Subcutaneous b uprenorphine injections were given prior to surgery and 24 h later at doses of 0. 05 mg/kg. Under deep anesthesia the sterile construct was implant ed
174 in to the peritoneal cavity via a small (2 cm) midline celiotomy. It was placed into the peritoneal cavit y and washed with saline containing 1% penicillin -streptomycin prior to closing. Two types of constructs surrounded by an implantable coil were studied: a cell -free implantable coil construct assembly and a cell containing implantable coil -construct assem bly. Cell -containing coil -construct assemblies were implanted in diabetic mice, whereas cell -free coil -construct assemblies were implanted in non diabetic mice. In addition, c ell -containing free floating beads were injected into the peritoneal cavity of diabetic mice via a syringe with a 16G needle under anesthesia for comparison studies. For diabetic mice, blood glucose measurements were taken daily to monitor the effectiveness of the construct s Once blood glucose levels exceeded 600 mg/dL or the mic e appeared distressed and weak they were euthanized and the construct was re trieved for histological analysis. For non-diabetic mice, magnetic resonance imaging (NMR) was performed from time to time. 6 3 9 Results and Statistical Analysis All GCR measurements and measurements performed on mice are indicated as the means standard deviation of multiple measurements derived from at least two independently encapsulated cultures. If a meaningful average could not be determined, the individual results are presented and commented. 6 4 Results 6 4 1 Implantable Coil -Construct Assembly Development 6 4 1 1 Implantable coila lginate foam assembly TC tet cells were successfully seeded in the alginate foam in the middle of the coated coi l The cells were growing in the foam and the cons tructs were kept alive for 56 days in vitro However, the cells grew slower than in a flask. They also kept falling out of the foam very rapidly. The alginate foam was later found to have high guluronic content which has already
175 shown cell grow th limita tion (175) Furthermore, the pores of alginate foams, even the 75% Ca2+ saturated ones (the least swelling foam type) were found to be too large to properly host cells within the construct. Regular autoclave sterilization methods did not seem the most appropria te for the alginate foam either. These concerns compromised th e use of alginate foam design in the scope of this study. If r educing alginate foam pore size could have been achieved by adding extra alginate gel layers around the foam t his implantable coil -construct assembly was not considered further in this study due to time constrains 6 4 1 2 Implantable c oil-a lgin ate b ead a ssembly TC tet cells were successfully encapsulated in alginate beads, either APA beads (Figure 6 3A) or BaA beads (Figure 6 3B) successfully inserted in the constructs immediately after generation. The con structs were then tested in vitro or in vivo in diabetic and non -diabetic mice. Figure 6 3. Photograph of TC tet cells encapsulated in alginate beads prior to insertion in a construct. A) APA beads; B) BaA beads. (Photo by N.A. Volland). 6 4 2 In v itro S tudy If originally the entire construct cavity was filled with beads (~ 0.3 mL), it was quickly observed that the cells did not survive very well in such an environment. Implantable c oil construct assemblies were placed in flasks containing free floatin g beads to determine if the 400
176 construct materials were an issue. No difference in bead -entrapped cell viability was observed between these two sets of free -floating beads leading to state that the construct materials had no effects on the cells. The restri cted movement of the beads due to their promiscuity when confined in the construct cavity compared to free floating bea ds was another possible issue. To increase the well -being of the encapsulated cells in beads, the volume of beads inserted in the constru ct was limit more freely within the construct cavity and the total number of cells presented in the construct cavity is still high enough to potentially reverse diabetes in later studies. Figure 6 4A i llustrates the temporal changes in the glucose consumption rate (GCR) of tet cells encapsulated in 2% (w/v) LVM alginate gelled either with 100 mM CaCl2 (APA beads) or 55 mM BaCl2 (BaA beads) and confined in constructs compared to beads free floating i n flasks (Figure 6 4B). Solid black square points ( ) correspond to data obtained from BaA bead culture in constructs (n = 6 constructs in 2 Petri dishes) whereas solid grey circle points ( ) correspond to data obtained from APA bead culture in constructs (n = 6 constructs in 2 Petri dishes). On Figure 4B, the solid black square points ( ) and solid grey circle points ( ) correspond to data obtained from beads cultured in flasks, APA beads and BaA beads respectively, used for comparison between restrained beads in constructs and free floating beads. The data show that both sets of cultures (APA and BaA beads) in constructs displayed an overall increase in glucose consumption comparable to the free -floating bead trend for both cultures as well. This incre ase in glucose consumption relates to an increase in cell proliferation as depicted by beads histology cross -sections (Figure 6 5 and 6 6).
177 Figure 6 4. Temporal changes in the glucose consumption rate (GCR) of TC tet cells encapsulated in APA and BaA beads A) When the beads where confined in coil construct assemblies; and B) when the beads where floating free in flasks. Solid square points ( ) correspond to the BaA beads and solid circle points ( ) correspond to the APA beads.
178 If three phases of glu cose consumption can be observed on Figure 6 4 for both types of beads (APA and BaA) in both environments (restrained or not), differences can be seen as well. In both environments, the cells in APA beads did not consume as much glucose early on as the ce lls in BaA beads. This pattern shows the influence of the manufacturing stress on the cell growth (longer generation procedure for APA beads). However, this trend reverses around Day 14 indicating the alginate network strength influence over time. It is also worthwhile mentioning that the alginate network strength may also influence the larger disparity in cell GCR for cells entrapped in APA beads only. Furthermore, cells in restrained beads in constructs seem to suffer more severe changes than the cell s in non -restrained beads. They consume more glucose immediately after insertion in the constructs compared to cells in free floating beads. Their consumption stabilizes later on before significantly dropping and increasing again. This pattern shows the influence of the construct confinement on the bead-entrapped cell growth. Nonetheless, this study shows that cells entrapped in alginate beads (APA and BaA) confined in construct can survive over a month in vitro The analysis of the histological cross -s ections also shows that the TC tet cells grew in clusters throughout the bead for both types of beads (APA and BaA) in both environments (restrained in construct or not). However, these clusters are more but smaller in BaA beads (Figure 6 5) than they ar e in the APA beads (Figure 66) in both environments. A similar observation was made earlier between APA beads made with alginates of different compositions (high mannuronic content vs. high guluronic content) (174,175) Furthermore, the clusters were slightly fewer and smaller for both types of beads when the beads were confined within a construct compared to when they were free floating in flask
179 Figure 6 5. Hematoxylin/eo sin stained cross -sections from BaA beads prepared by encapsulating TC tet cells in 2% (w/v) LVM alginate gelled with 55 mM BaCl2. A) At Day 0, B) at Day 36 after retrieval from the construct; and C) at Day 30 after retrieval from the flask. (Photo by N .A. Volland). Figure 6 6. Hematoxylin/eosin stained cross -sections from APA beads prepared by encapsulating TC tet cells in 2% (w/v) LVM alginate gelled with 100 mM CaCl2. A) At Day 0, B) at Day 30 after retrieval from the construct; and C) at Day 30 after retrieval from the flask. (Photo by N.A. Volland).
180 These histological observations corroborate well with the differences in glucose consumption displayed between the two beads types in both environments (restrained in construct or free floating). This indicates that BaA bead and construct environment both hindered the cell growth. Although only one bead histology cross -section photograph is presented here for each type of beads per environment. It is also worthwhile mentioning that the APA bead histology cross -sections did display a larger discrepancy among themselves than the BaA m ade cross -sections as larger discrepancies were since in the GCR graphs 6 4 3 In vivo S tudy Two in vivo studies were performed: 1) the implantation of cell -free implantable coil construct assemblies and cell -free coil -free constructs in non diabetic mice to study the effect of the assemblies on the animal and vice versa and 2) the implantation of cell -containing implantable -coil -construct assemblies in diabetic mice to asses s the efficacy of the construct and determine possible issues and/or improvement s Cell -free implantable coil -construct assemblies and cell -free coil -free constructs were kept in mice peritoneal cavities up to 123 days without apparent external problem for the animals (n = 14) with an average of 40 days. None of the implantable coil s ever failed (n = 7) NMR images were taken at any given day on any mouse without a problem. Every assembly implanted for 4 days or less was perfectly clean upon retrieval. However, every construct assemblies but two (n = 9) which stayed implanted for more than 4 days had a thin ( 7A) around them even when antibiotics such as penicillin and streptomycin, and the steroid dexamethasone were us ed to reduce the immune response 58% of the se assemblies had the construct cavity filled up with more fibrous connective tissue (Figure 6 7B). This second fibrous tissue layer only seems to develop around the alginate beads and never attack them directl y since the beads have always been appeared intact on histology cross -sections (Figure 6 7B).
181 Figure 6 7. Hematoxylin/eosin stained cross -sections of the tissues found inside and around the construct after its retrieval from a mouse peritoneal cavity A) F ibrotic tissue layer piece found around the construct at the time of retrieval; and B) Connective tissue present inside the construct cavity at the time of retrieval and surrounding the cell free alginate beads inserted in the cavity before implanta tion. (Photo by N.A. Volland). If the fibrous connective tissue layer inside the assembly cavity was often seen on the MR images (Figure 6 8), the thin fibrotic layer around the construct was never observable. It was collected at the time of construct re trieval. It was never attached onto the PDMS part of the construct, but onto the mesh. Figure 6 8. MR images of pancreatic substitutes implanted in the peritoneal cavity of two different mice. The images show the presence of connective tissue within t he construct cavity after: A) 39 days of implantation ; and B) 98 days of implantation ~ 200 m
182 Changes in construct irrigation could be perceived on the MR images as a result of this fibro tic encapsulation of the whole construct since the beads appeared more disti nctively within the cavity over time as the construct water content decreased (Figure 6 9). Figure 6 9. MR images of cell -free beads within the construct cavity over time. A) At day 10; and B) at day 42. Upon these results, mice were implanted with ei ther a piece of PEEK mesh only or a ring of PDMS over one week. If the PDMS rings were retrieved perfectly clean, the PEEK mesh was well covered (Figure 6 10B). Figure 6 10. Photographs of a piece of PEEK mesh. A) prior to; and B) after implantation in a mouse peritoneal cavity for a week. (Photo by N.A. Volland).
183 T wo diabetic mice were also implanted with coil-construct assemblies containing a volume TC tet cells entrapped in alginate beads gelled with 55 mM BaCl2 (n = 2). Two other d for comparison purposes The mice became diabetic within 10 to 15 days of the last STZ injection and were all successfully implanted (Day 14) or injected (Day 18). Figure 6 11A illustrates the temporal c hanges in the blood glucose levels of diabetic mice implanted with constructs containing BaA beads. Solid grey square points ( ) and solid black circle points ( ) correspond to data obtained from the mice which received the coil -construct assembly contain ing BaA beads. In figure 611B, Solid grey square points ( ) and solid black circle points ( ) correspond to data obtained from the mice which received the free floating BaA bead injection. These last mice were used for comparison between the mice receiv ing the restrained beads in constructs and the ones receiving the free floating one. Blood glucose levels of normal mice (nave controls, solid light grey rhombus points ( ) on Figure 6 11) and untreated diabetic mice (negative controls) were also checked to give a reference to how well the constructs and beads were performing. No averaging was done in this experiment to generalize the findings obtained since the number of animals tested was very small (up to 2 animals per set) and the results would have been biased by the large disparity from one animal to another. However, encouraging observations can be made.
184 Figure 6 11. Temporal changes in diabetic mice blood glucose levels after implant ation of TC tet cells encapsulated in BaA beads. A) A s par t of a coil -construct assembly; and B) free floating in the peritoneal cavity Solid square s ( ) and solid circles ( ) correspond to two different mice which received the beads either confined in construct (A) or through injection (B). Solid rhombus points ( ) correspond to the nave control mice. The dashed grey lines placed at 150 and 250 mg/ml set the average blood glucose level range for control mice. The solid black line placed at 300 mg/ml set the blood glucose level above which mice are considere d diabetic after their blood glucose is detected 3 days above this level The solid black line placed at 600 mg/ml set the blood glucose level above which mice have to be sacrificed after their blood glucose is detected 3 days above this level.
185 The data s how that the two mice which received the constructs return to normoglycemia within 24 or 48 hours of implantation indicated that the constructs are playing their role of pancreatic substitutes in producing insulin to compensate the lack of this hormone due to diabetes. Despite the fact that free floating beads restored normoglycemia for a longer period (up to 10 days) than coil -construct assembly beads, normoglycemia was restored for between 6 and 12 days in diabetic mice which received the constructs. The mice were considered diabetic again 13 days after implantation. It is worthwhile noting that it took more than 10 days for the blood glucose to reach the pre implantation levels again. If the constructs were retrievable for histology studies, none of f ree floating beads could ever be retrieved for comparison This shows the advantage of a pancreatic construct assembly to help toward the development of a practical bioartificial pancreas in the treatment of diabetes. Figure 6 12 displays a histological cross -section photograph of a piece of tissue including a cell -containing bead obtained from inside the construct cavity of one of the mice implanted with a coil -construct assembly The construct was retriev ed on Day 35 of the study or 21 days after impla ntation Figure 6 12. Hematoxylin/eosin stained cross -section from implantable coil -construct assembly cavity tissue after 21 days of implantation in a mouse peritoneal cavity. The cavity included beads prepared by encapsulating TC -tet cells in 2% (w/ v) LVM alginate, gelled with 55 mM BaCl2. (Photo by N.A. Volland).
186 Cells can be spotted inside the bead present on the cross -section photograph (Figure 6 12). However, their number seems closer to the initial number of cells ( 3.5 x107 cells/ml of alginat e, Figure 6 5A) than to the number of cells at the end of the in vitro study (Figure 6 5B). Fibrous c onnective tissue can also been seen in the cavity outside of the beads attesting of an immune response toward the construct. 6 5 Discussion Alginate beads hav e previously been shown to restore normoglycemia in mice (35,177,178) over a period of 5 d ays. However, their dispersion throughout the hosts body limits obtaining valuable information on structure, metabolic activity, or function of cells entrapped in the beads and optimizing such a construct. To overcome this limitation, the use of a macro construct was investigated and characterized in vitro and in vivo Macroconstructs have the following advantages over microconstructs, such as beads: 1) they can be held in place in the host peritoneal cavity, and 2) they can be surrounded by an RF coil ( implantable coil -construct assembly), so that its monitoring can be done performing NMR imaging and spectroscopy. Data in this study show s that a n implantable coil -construct assembly including alginate beads can allow insulin-producing cells to stay viable for an extended period of time in vitro T he growth of the bead -entrapped cells restrained in a construct may not have been as steady as when the beads were not restrained. However, these results shows that cells can be entrapped in beads restrained in construct and their number increased to compensate for their performance limitation due to their environment restriction s. This allows the bioartificial pancreas to be noninvasively monitored in situ If APA beads have been generated routinely in our la boratory (151,175) BaA bead generation was implemented in an effort to accelerate the bead -making process with less washes not to leave the cells in a harsh er environment than normal for an extended period of time. Figure 6 4 attested of the slower cell growth early on in APA beads
187 due to the length of their generation process and its potential to be more harmful than the BaA bead generation process As the ionic binding between alginate chain and barium cations is stronger than the binding between alginate chain and calcium cations (59) no PLL coating and additional washes were necessary to shape the BaA beads and give them the desired porosity. The histology cross -sections attested of th e stronger network provided by the BaA beads than the network provided by the APA beads as the cell clusters are smaller in BaA beads than in APA beads From the previous studies between APA beads made with alginates of different compositions ( high mannur onic content vs. high guluronic content) (1 74,175) the softer alginate network was often favored. However, the BaA beads may be more appropriate to use in a construct in vivo The fact that no PLL is used to make BaA beads may help decrease the host immune response and ease protein assay tests upon retrieval such as the Lowry protein assay to determine cell numbers and growth with accuracy. Furthermore, with a tighter network, beads are holding together better and longer, allowing the construct to last longer and generate more reproducible and consistent results. O ur laboratory is currently investigating the optimal strength of alginate bead network for cell growth when cells are entrapped in beads confined in a construct. Generating beads using different concentration of barium cations is be ing testing for this matter. It is worthwhile noting that no tetracycline was used in the in vitro studies to regulate the tet cells and that the study was stopped at 30 days before the beads broke apart due to the large number of cells in them. Adding tetracycline in the culture media will prevent cell growth and allow the studies to go on for a longer time more appropriate for future in vivo studies (151)
188 The impla ntable coil -construct assembly implantation in a mouse peritoneal cavity demonstrates the capacity of cells entrapped in alginate beads confined in construct to detect glucose and release insulin to control the host blood glucose levels while allowing for enhanced non invasive monitoring of the construct using NMR methods. The fact that every implantable coil implanted never failed over the course of the in vivo studies shows how steady and reliable these coils are. These results also showed that the cel ls were not impaired by coils and that the return to normoglycemia in diabetic mice wa s possible upon coil construct assembly implantation. T he cells were shown to produc e enough insulin to keep the mice blood glucose levels below the diabetic range for no more than 5 consecutive days. U pon retrieval, most of the construct cavities were filled with extracellular tissue and surrounded by a fibrotic layer typical of a host immune response Assembly pre -treatment s decreased th is host immune response but di d not eliminate it completely. Nonetheless, it is worthwhile noting that 10 days were necessary to see the mice blood glucose levels back to their preimplantation levels. T his study will need to be repeated over a larger number of animals to obtain bette r statistics as the differences between animals can be wide. Furthermore STZ solutions were found to be very instable and hard to properly implement all the time to render the mice diabetic in a reasonable number of days (5 10 days). The type of mice us ed in this study was also found to be mildly responsive to STZ. IP Alloxan injections are currently under investigation in our laboratory to increase the efficiency in rendering mice diabetic. Further studies of the host immune response on the pancreatic construct found that the PEEK mesh triggered a large reaction of the host and its rapid coverage by host fibrotic cell layer (within less than 10 days) even though biocompatible, protein -free, and sterile. This reaction could explain the filling of the construct cavity with connective tissue, the fibrotic layer
189 appearance outside the construct, and the rapid decrease of construct performances over time. If the connective tissue in the construct cavity may not have kill ed the cells directly as it does not appear densely cellularized, its cells certainly compete with the insulin -producing cells for nutrient and oxygen. On the contrary, t he fibrotic layer outside the construct does appear very dense and could be responsible for the cell oxygen depravation and lack of fluid irrigation. All of these assumptions are possible schemas of rapid construct failure which NMR methods have already been able to partially identify. The mesh composition, roughness and topography have been mentioned as possible reasons for triggering a response from the host immune system. Studies in our laboratory currently investigate the coating of the mesh with pure alginate to alleviate the issues and allow the construct to perform optimally for an extended period of time. The use o f a different type of mesh may also help decreasing the immune response and will need to be investigated in the future 6 6 Summary The implantable coil -construct assemblies were successfully developed and their performances analyzed. T he increased cell viabi lity when using microencapsulation and the cell confinement when using macroencapsulation was c ombin ed to: 1) allow the generation of a bioartificial pancreas with higher diffusion capability; and 2) keep the cells together in one location to monitor them closely and easily in vivo using NMR techniques and implantable coil s The cells entrapped in alginate beads confined inside the construct cavity were maintained alive for over a month in vitro This could be extended in the future with the use of tetrac ycline to regulate cell growth. Additionally these assemblies show their potential to restore normoglycemia in diabetic mice Some of the immune response issues which prevented the construct to function properly more than 5 consecutive days have already been isolated. The PEEK mesh used to create the construct cavity to hold the beads in place was shown to trigger a
190 large immune response within 10 days a fter implantation. C onstruct pretreatment in parallel to the use of antibiotic were found to slightl y decrease the fibrotic overgrowth around the construct. However the close NMR noninvasive monitoring possible because of the implantable RF coil around the construct showed how the MR images could allow the identification of such a host reaction prior to the return of hyperglycemia It is worthwhile noting that the coil function w as never impaired by this host immune response. Since any immune response will impaired the construct function within days after implantation, this issue needs to be addressed before the bioartificial pancreas performances can be fully assessed Consequently, future investigations should focus on modifying the construct to decrease, and even eliminate, the host immune response to maintain the cells alive longer and allow the c onstruct to restore normoglycemia in diabetic mice ov er an extended period of time.
191 CHAPTER 7 CONCLUSIONS AND FUTU RE WORK 7 1 Conclusions and Contributions A sensitive and noninvasive monitoring is of critical importance for a complete and accurate characterization o f tissue -engineered constructs in vitro and in vivo and the early detection of construct failure markers. The primary contribution of this work is the sensitivity improv ement of an existing noninvasive NMR monitoring method by developing inductivelycoup led implantable coil system for use at high field This is the first time that the use of inductively-coupled coil systems was implemented at a magnetic field as high as 11.1T. While the sensitivity improvement was essential the ease of coil system cons truction and the implantable coil coating procedure were all taken into account to decide the most advantageous design overall. Simulations, phantom, and in vivo studies helped to characterize the coil systems and provided valuable information regarding t he changes in coil resonant frequency, quality factor, and inductive coupling, under different loading conditions. Different designs (loop -gap resonators and solenoids) and configurations (co and counter rotating modes) were evaluated T he use of induct ively -coupled implantable coil system in the co rotating mode allowed the high est sensitivity improvement of the NMR monitoring method for 1H detection The loop -gap resonator was selected as the system implantable coil because of easy construction and re producibility. The coating of the implantable coil was a significant pa rt of the system development and t he effects of coating on the implantable coil and the whole system were determine d for different materials. PDMS was selected to be the most appropri ate coating material and allow the use o f the coupled -coil system in vitro and in vivo with the least sensitivity sacrifices The design of the best possible inductively-coupled implantable coil system for the model tissue -engineered construct and NMR app lication s specific to this research
192 was achieved T he developed coupled coil system with a PDMS -coated loop gap resonator allowed a sensitivity improvement of a non invasive NMR monitoring technique at 11.1 T for 1H detection in vivo over 4 times when com pared to the same technique using a surface coil at 4.7 T The second phase of the research upgraded the inductively-coupled implantable coil system to receive-only and multiple -frequency coupled -coil systems. R eceive -only systems were successfully buil t using passive decoupling circuit Design and coating adaptations from the inductively-coupled system construction for 1H detection were effectively implemented validating that more complex coupled -coil systems are feasible NMR images and spectra were also obtained using receive -only coupled-coil systems at 11.1 T demonstrating the capability of these system s to detect NMR signal with a high sensitivity T he excitation homogeneity and increased selectivity available us ing the receive -only inductively -c oupled implantable coil system allowed the acquisition of spectroscop ic data with greater SNR and better localization. The development of multiple -frequency coupled -coil system for 1H and 31P and 1H and 19F simultaneous detection were also investigated to assess their feasibility. Single -frequency systems for 19F and 31P detection w ere constructed and indicated the possibility of detecting several different relevant nuclei in the monitoring of a bioartificial pancreas The single frequency inductivelyco upled implantable coil system s also gave the reference characteristics for the multiple -frequency inductively-coupled implantable coil systems. The use of tank circuit s and capacitive overcoupling showed the feasibility of multiple -frequency coupled -coil systems. M atching issues, resistive losses, and unexpected frequency coupling within the double frequency systems indicated how complex these systems can become and that the development require s extensive considerations prior to magnet testing However, the technology of inductively-coupled implantable coil system s even the complex ones, has been established and
193 allow ed the monitoring of a bioartificial pancreas and the detection of early marker s of construct failure using NMR methods with a high sensiti vity. The final phase of this dissertation tested the pancreatic construct characteristic when inserted in the middle of the implantable coil. This was the first time an implantable coil surrounded a pancreatic construct for the direct monitoring of its f unction. The design of a hybrid micro/macroconstruct using a ring of polydimethylsiloxane (PDMS) encasing alginate beads and an implantable coil proved that entrapped cells can be kept alive in vitro for over a month and can restore normoglycemia in diabe tic mice for 5 consecutive days These results showed that cells can survive, grow, and function properly when confined within a pancreatic construct including an implantable RF coil. The results also demonstrated the ability of the inductively-coupled i mplantable coil system to directly monitor a viable bioartificial construct in vivo with high sensitive using NMR methods. Pre -treating the construct prior to implantation was found to be of primary importance to limit construct fibrotic overgrowth and pr eserve the longevity of the construct in vivo as much as possible The data presented in this dissertation outline s the quantitative criteria for the capability and limitations, such as coil design and coating materials, of using inductively-coupled implan table coil system s with an NMR method for monitoring a bioartificial pancreas non invasively. The use of inductively-coupled implantable coil system, under the specific conditions used in this study, was found to improve the sensitivity of a noninvasive NMR monitoring technique for 1H detection by a factor of 2 compared to the same technique when using a surface coil (SC) at 11.1 T. This was added to the sensitivity improvement of 2.4 when going to a higher magnetic field than the previous studies (from 4.7 to 11.1 T). In addition, these inductively-coupled implanted coil systems will have larger advantages over an SC as the depth
194 of the implanted coil increases, such as in larger animal models (e.g. humans), since the SC sensitivity drops rapidly with distance. While the implantable coil design discussed here appear to limit the scope of the NMR non-invasive monitoring to bioartificial pancreatic construct, i t is worth while mentioning that the coil developments outlined in this thesis can also be applie d to monitor other systems tissue and/or organs bioartificial or not, with NMR methods In these cases, the coil geometry may have to be adapted to better fit the studied structure shape and size. Moreover, this increase in sensitivity has driven the development of more complex systems, such as receive-only and multiple -frequency inductively -coupled implantable coil systems. The receive-only coupled -coil systems w ere investigated to allow a better localization within the construct to perform spectroscopy and the multiple -frequency coupled -coil systems to detect important nuclei other than 1H. Their development has show n the potential for higher sen si tivity localized spectroscopy and the creation of a unique implantable coil system for the simultaneous assessment of different physiologically relevant parameters within a single tissue -engineered construct 7 2 Future Work The data presented in this thesis were primarily relegated to developing inductively coupled implantable coil system to improve the sensiti vity of an NMR non invasive monitoring method. In order to increase the sensitivity even further different designs, different coating materials, and different diodes need to be tested. Using a coating material with better dielectric properties than PDMS would further limit the electrical losses encountered during the coating and loading of the coil. PTFE, with its excellent dielectric properties, was tested without success, but reconsidering this material with more adequate coating process may work and h elp better isolate the implantable coil from its surrounding. However, caution will have to be taken when 19F detection is performed. Signal suppression schema may have to be applied if PTFE
195 signal is detectable by NMR. If this signal interferes with the PFC signal of interest in this study, other materials, which are non -detectable by NMR, should be conside red or even developed, such as p arylene C (1 32) or polyurethane (12) Even though the receive -only inductively coupled coil system was successfully tested in the magnet, the use of diodes with more favorable characteristics would minimally alter the coils and inductively -coupled system characteristics ease the system construction, and potentially increase the sensitivity of NMR method sensitivity when the receive only system is used The non -magnetic passively activated Schottky diode used in this study had a series resistance superior to 1 a parallel capacitance superior to 3 pF, and a parallel resistance averaging 5 k These characteristics largely influence both the surface and implantable coils because the diode parallel capacitance is of the order of magnitude of the circuit capacitance during NMR signal reception when the diode is reverse biased, or OFF. A diode wi th a smaller parallel capacitance ( the diode is OFF and acting as a capacitance, it has a specified parallel resistance indicative of the component quality factor The larger this resistance, the least influence the decoupling circuit will have on the main circuit when the diode is OFF. Moreover, d uring NMR excitation, the diode is forward biased, or ON, and equivalent to a series resistance. The smaller this re sistance (< 1 ), the better decoupled the coils will be since the current will mostly flow in the decoupling circuit and not the main circuit. PIN diodes can achieve series resistance inferior to 1 parallel capacitance inferior 1 pF, and parallel resi stance superior to 25 k to minimally influence the system they are inserted in, but require external DC power to achieve such characteristics The passive activation required in these experiments does not allow the use of actively driven PIN diodes. T he development of non-magnetic diode s with lower series
196 resistance and parallel capacitance when passively driven may be worth while if not found commercial ly The interference of breathing artifacts on the acquisition of NMR data is another aspect of this r esearch that would benefit from future studies (179) Although the use of respiratory gating in this thesis was able t o effectively reduce the artifacts to the noise level, this technique could not be accurately implemented when the mice were breathing irregularly, which could happen frequently when the mice are diabetic and weak, and/or under low SNR conditions (low magn etic field strength and clinical setting) Different strategies can be implemented to eliminate th e se breathing artifacts. If self -gating is widely used in cardiac MR imaging (155157) a user -defined gating technique looking more promising as the acquisition time of standard MR images is longer (158) However, an image reconstruction using a motion artifact correction post processing could enhance the image quality even furthe r. Different techniques have already been developed for abdominal imaging in clinical settings and may be implemented in the frame of this research to reduce breathing artifact when imaging the construct implanted in a mouse peritoneal cavity (180,181) Another aspect for future studies would be the improvement of the multiple -freq uency inductively-coupled implantable coil system. Future work in this area should focus on two aspects: (1) tun ing a multiple -frequency inductively -coupled implantable coil system at two frequencies (1H -31P & 1H -19F) and match it to 50 and (2) tun ing a multiple -frequency inductively-coupled implantable coil system at three frequencies (1H -31P -19F) and match to 50 as well. Either of these approaches would result in the system ability to allow direct cross correlation studies between the different cel l metabolic activities and functions. Obtaining a better matching networking for the 1H -31P system using adequate variable inductors can be
197 investigated. Different system designs have to be studied to obtain a working 1H -19F system without 1H and 19F fre quencies interacting with each other to test these systems in the magnet. However, b ecause adding resonances into a coil circuit generate more and more resistive losses at any given resonant frequency of the system, an alternative multiple -frequency induc tively coupled implantable coil system using a single resonance approach is under development in our laboratory (182). A microcontroller embedded within a microchip and integrated to the coils can switch an array of capacitors to make the coil resonate at the desired frequency when remotely activated. The arrays have been shown to allow the tuning of the coils at any frequency making the system multiple -frequency with a single frequency system behavior. The techniques demonstrated in this dissertation were intended to be used in total choline signal measurement for construct cell viability determination, ATP signal me asurement for construct bioenergetics status determination, and PFCE T1 relaxation time measurement to assess the oxygenation of the construct over time. Nonetheless, all those measurements need to be carefully calibrated for the use of induc tively couple d implantable coil system before studying a construct over time. Total -choline and ATP measurements are relatively easy to calibrate as they only require using sample of different concentration of the desired compound (known cell density) to be studied. However, PFCE T1 measurement calibration will require the construction of a perfusion system where the system oxygenation is known and can be carefully regulated. All these measurements would provide a large step forward in the eventual realization of pra ctical, clinically relevant constructs. Furthermore, total -choline signal has yet to be acquired in vitro at 11.1T before the calibration and longitudinal studies of the construct can be done. The shim procedure can be improved by manually shimming of th e second order shims or optimizing the FASTMAP for the voxel size required in the study. Pulse sequences can also be optimized or
198 changed to a stimulated echo acquisition mode (STEAM) pulse sequence to take into account the possibility of a relatively sho rt T2 (83) R eproducing the previous study at 4.7 T may also be a good starting point before going to higher field and more complex coil systems No extensive construct viability and performance in vivo studies were performed in the scope of this research as the host immune response was impairing the construct function within days after implantation. However, t his immune response problem triggered preliminary studies as it was a major issue to be addressed before going any fur ther with any construct viability and performances studies. Besides pretreating the construct prior to implantation and using antibiotics, c oating the coil -construct assembly with a more bioinert material, such as ultra pure alginate is currently under i nvestigation in our laboratory. The use of a different mesh material h as also been brought up and will be considered upon the alginate overcoating results to decrease, and even eliminate, the host immune response, maintain the cells alive longer and allow the construct to restore normoglycemia in diabetic mice over an extended period of time.
199 LIST OF REFERENCES 1. Silva AI, de Matos AN, Brons IG, Mateus M. An overview on the development of a bioartificial pancreas as a treatment of insulin -dependent diabetes mellitus. Medicinal Research Reviews 2006;26(2):181222. 2. Constantinidis I, Long RCJ, Weber C, Safley S, Sambanis A. Noninvasive monitoring of a bioartificial pancreas in vitro and in vivo. Annals of the New York Academy of Sci ences 2001;944:8395. 3. Constantinidis I, Sambanis A. Noninvasive monitoring of tissue engineered constructs by nuclear magnetic resonance methodologies. Tissue Engineering 1998;4(1):9 17. 4. Constantinidis I, Stabler CL, Long RCJ, Sambanis A. Noninvasive monitoring of a retrievable bioartificial pancreas in vivo. Annals of the New York Academy of Sciences 2002;961:298301. 5. Stabler CL, Long RCJ, Constantinidis I, Sambanis A. Noninvasive measurement of viable cell number in tissue engineered constructs i n vitro using 1H NMR spectroscopy. Tissue Engineering 2005;11(34):404414. 6. Stabler CL, Long RCJ, Constantinidis I, Sambanis A. In vivo noninvasive monitoring of viable cell number in tissue engineered constructs using 1H NMR spectroscopy. Cell Transpla ntation 2005;14(23):139149. 7. Hoult DI, Lauterbur PC. The sensitivity of the zeugmatographic experiment involving human samples. Journal of Magnetic Resonance 1979;34:425433. 8. Haase A, Odoj F, von Kienlin M, Warnking J, Fidler F, Weisser A, Nikita M, Rommel E, Lanz T, Kalusche B, Griswold M. NMR probeheads for in vivo applications. Concepts in Magnetic Resonance 2000;12:361388. 9. Koretsky AP, Wang S, Murphy-Boesch J, Klein MP, James TL, Weiner MW. 31P NMR spectroscopy of rat organs, in situ using c hronically implanted radiofrequency coils. Proceedings of the National Academy of Sciences USA: Biochemistry 1983;80:74917495. 10. Murphy-Boesch J, Koretsky AP. An in vivo NMR probe circuit for improved sensitivity. Journal of Magnetic Resonance 1983;54:526532. 11. Hollett MD, Cofer GP, Johnson GA. In situ magnetic resonance microscopy. Investigative Radiology 1987;22:965968. 12. Schnall MD, Barlow C, Subramanian VH, Leigh JSJ. Wireless implanted magnetic resonance probes for in vivo NMR. Journal of Magnetic Resonance 1986;68:161167.
200 13. Zhang X, Zhu XH, Tian R, Zhang Y, Merkle H, Chen W. Measurement of arterial input function of 17O water tracer in rat carotid artery by using a region -defined (REDE) implanted vascular RF coil. MAGMA 2003;16(2):7785. 14. Silver X, Ni WX, Mercer EV, Beck BL, Bossart EL, Inglis B, Mareci TH. In vivo 1H magnetic resonance imaging and spectroscopy of the rat spinal cord using an inductivelycoupled chronically implanted RF coil. Magnetic Resonance in Medicine 2001;46:12161222. 15. Bilgen M. Simple, low cost multipurpose RF coil for MR microscopy at 9.4T. Magnetic Resonance in Medicine 2004;52:937940. 16. Bilgen M. Inductively -overcoupled coil design for high resolution magnetic resonance imaging. Biomedical Engineering onli ne 2006;5(3). 17. Bilgen M, Al Hafez B, He Y Y, Brooks WM. Magnetic resonance angiography of rat spinal cord at 9.4T: a feasibility study. Magnetic Resonance In Medicine 2005;53:14591461. 18. Bilgen M, Elshafiey I, Narayana PA. In vivo magnetic resonance microscopy of rat spinal cord at 7 T using implantable RF coils. Magnetic Resonance in Medicine 2001;46(6):12501253. 19. Fenyes DA, Narayana PA. In vivo echo -planar imaging of rat spinal cord. Magnetic Resonance Imaging 1998;16(10):12491255. 20. Narayana PA, Fenyes D, Zacharopoulos N. In vivo relaxation times of gray matter and white matter in spinal cord. Magnetic Resonance Imaging 1999;17(4):623626. 21. Wirth EDI, Mareci TH, Beck BL, Fitzsimmons JR, Reier PJ. A comparison of an inductively coupled impl anted coil with optimized surface coils for in vivo NMR imaging of the spinal cord. Magnetic Resonance in Medicine 1993;30:626633. 22. National Diabetes Information Clearinghouse. National Diabetes Statistics, 2007. National Institute of Diabetes and Dige stive and Kidney Diseases E d: National Institute of Health; 2007. 23. Yoon J -W, Jun H -S. Autoimmune destruction of pancreatic -cells. American Journal of Therapeutics 2005;12:580591. 24. National Diabetes Information Clearinghouse. Diabetes Type I. Nati onal Institute for Diabetes and Digestive and Kidney Diseases ed. Washington, D.C.: National Institute of Health; 2008. 25. Robertson RP. Pancreatic and islet transplantation for diabetes -cures or curiosities. New England Journal of Medicine 1992;327:18611868.
201 26. Sutherland DER. Pancreas and islet cell transplantation: now and then. Transplantation Proceedings 1996;28:21312133. 27. Shapiro AM, Lakey JRT, Ryan E, Korbutt G, Toth E, Warnock GL, Kneteman NM, Rajotte RV. Islet transplantation in seven patien ts with type I diabetes mellitus using a glucocorticoid -free immunosuppressive regimen. New England Journal of Medicine 2000;343:230238. 28. Shapiro AM, Lakey JR, Ryan EA, Korbutt GS, Toth E, Warnock GL, Kneteman NM, Rajotte RV. Islet transplantation in s even patients with type 1 diabetes mellitus using a glucocorticoid -free immunosuppressive regimen. New England Journal of Medicine 2000;343(4):230238. 29. Sun AM, Parisius W, Macmorine H, Sefton MV, Stone R. An artificial pancreas containing cultured isle ts of Langerhans. Artificial Organs 1980;4:275278. 30. Sullivan SJ, Maki T, Borland KM, Mahoney MD, Solomon BA, Muller TE, Monaco AP, Chick WL. Biohybrid artificial pancreas: long term implantation studies in diabetic, pancreatectomized dogs. Science 1991;252(5006):718721. 31. Lembert N, Peterson P, Wesche J, Zschocke P, Enderle A, Doser M, Plank H, Becker HD, Ammon HPT. In vitro test of new materials for the development of a bioartificial pancreas. Annals of the New York Academy of Sciences 2001;944(1):271 276. 32. Sun AM, Parisius W, Healy GM, Vacek I, Macmorine H. The use in diabetic rats and monkeys, of artificial capillary units containing cultured islets of Langerhans (artificial endocrine pancreas). Diabetes 1977;26:11361139. 33. Sun AM, O'Shea GM. Microencapsulation of living cells: a long-term delivery system. Journal of Controlled Release 1985;2:137141. 34. Lim F, Sun AM. Microencapsulated islets as bioartificial endocrine pancreas. Science 1980;210:908910. 35. SoonShiong P, Feldman E, Nelson R, Komtebedde J, Smidsrd O, Skjk-Brk G, Espevik T, Heintz R, Lee M. Successful reversal of spontaneous diabetes in dogs by intraperitoneal microencapsulated islets. Transplantation 1992;54:769774. 36. Sakai S, Ono T, Ijima H, Kawakami K. In vitro and i n vivo evaluation of alginate/sol -gel synthesized aminopropyl -silicate/alginate membrane for bioartificial pancreas. Biomaterials 2002;23:41774183. 37. Benson JP, Papas KK, Constantinidis I, Sambanis A. Towards the development of a bioartificial pancreas: effects of poly -L lysine on alginate beads with TC3 cells. Cell Transplantation 1997;6:395402. 38. Lanza RP, Sullivan SJ, Chick WL. Islet transplantation with immunoisolation. Diabetes 1992;41:15031510.
202 39. Hirotani S, Ohgawara H, Agishi T, Akaike T, M iyazaki S. A bio artificial endocrine pancreas for the treatment of diabetes. Transplantation Proceedings 1998;30:485489. 40. Ohgawara H, Miyazaki J, Karibe S, Tashiro F, Akaike T, Hashimoto Y. Embedded culture of pancreatic -cells derived from transgeni c mouse insulinoma as a potential source for xenotransplantation using a diffusion chamber. Cell Transplantation 1995;4:307313. 41. Qi M, Gu Y, Sakata N, Kim D, Shirouzu Y, Yamamoto C, Hiura A, Sumi S, Inoue K. PVA hydrogel sheet macroencapsulation for th e bioartificial pancreas. Biomaterials 2004;25:58855892. 42. Halban PA, Wollheim CB. Intracellular degradation of insulin stores by pancreatic islets in vitro. An alternative pathway for homeostasis of pancreatic insulin content. Journal of Biological Che mistry 1980;255 (13):60036006. 43. Efrat S, Linde S, Kofod H, Spector D, Delannoy M, Grant S, Hanahan D, Baekkeskov S. Beta -cell lines derived from transgenic mice expressing a hybrid insulin gene -oncogene. Proceedings of the National Academy of Sciences USA: Cell Biology 1988;85:90379041. 44. Miyazaki J I, Araki K, Yamato E, Ikegami H, Asano T, Shibasaki Y, Oka Y, Yamamura K I. Establishment of a pancreatic cell line that retains glucose -inducible insulin secretion: special reference to expression of g lucose transporter isoforms. Endocrinology 1990;127(1):126132. 45. Dionne KE, Colton CK, Yarmush ML. Effect of hypoxia in insulin secretion by isolated rat and canine islets lf Langerhans. Diabetes 1993;42(1):1221. 46. Dionne KE, Colton CK, Yarmush ML. E ffect of oxygen on isolated pancreatic tissue. ASAIO transactions 1989;35(3):739741. 47. Papas KK, Long RCJ, Constantinidis I, Sambanis A. Effects of oxygen on the metabolic and secretory activities of bTC3 monolayers. Biochimica Biophysica Acta 1996;1291:163166. 48. Efrat S, Fusco DeMane D, Lemberg H, Emran OA, Wang X. Conditional transformation of a pancreatic -cell line derived from transgenic mice expressing a tetracycline regulated oncogene. Proceedings of the National Academy of Sciences USA: Bioch emistry 1995;92:35763580. 49. Thule PM, Liu J, Phillips LS. Glucose regulated production of human insulin in rat hepatocytes. Gene Therapy 2000;7:205214. 50. Tang S -C, Sambanis A. Development of genetically engineered human intestinal cells for regulated insulin secretion using rAAV -mediated gene transfer. Biochemical and Biophysical Research Communications 2003;303:645652.
203 51. Cheung AT, Dayanandan B, Lewis JT, Korbutt GS, Rajotte RV, Bryer -Ash M, Boylan MO, Wolfe MM, Kieffer TJ. Glucose dependant insul in release from genetically engineered K cells. Science 2000;290:19591962. 52. Lumelsky N, Blondel O, Laeng P, Velasco I, Ravin R, Mckay R. Differentiation of embryonic stem cells to insulin-secreting structures similar to pancreatic islets. Science 2001; 292:13891394. 53. Peck AB, Cornelius JG, Schatz D, Ramiya VK. Generation of islets of Langerhans from adult pancreatic stem cells. Journal of Hepatobiliary and Pancreatic Surgery 2002;9:704709. 54. Bonner -Weir S, Sharma A. Pancreatic stem cells. Journal of Pathology 2002;197:519526. 55. Kuo CK, Ma PX. Ionically crosslinked alginate hydrogels as scaffolds for tissue engineering: part 1. Structure, gelation rate, and mechanical properties. Biomaterials 2001;22:511521. 56. Hirotani S, Eda R, Kawabata T, F uchinoue S, Teraoka S, Agishi T, Ohgawara H. Bioartificial endocrine pancreas (Bio -AEP) for treatment of diabetes: effect of implantation of Bio -AEP on the pancreas. Cell Transplantation 1999;8:399404. 57. Lahooti S, Sefton MV. Agarose enhances the viabil ity of intraperitoneally implanted microencapsulated L929 fibroblasts. Cell Transplantation 2000;9:785796. 58. Martinsen A, Skjk Brk G, Smidsrd O. Alginate as immobilization material: I. Correlation between chemical and physical properties of alginate gel beads. Biotechnology and Bioengineering 1989;33(1):79 89. 59. Mrch YA, Donati I, Strand BL, Skjk -Brk G. Effect of Ca2+, Ba2+, and Sr2+ on Alginate microbeads. Biomacromolecules 2006;7:14711480. 60. Martinsen A, Storr I, Skjrk Brk G. Alginate a s immobilization material: III. Diffusional properties. Biotechnology and Bioengineering 1992;39(2):186194. 61. Smidsrd O, Skjk Brk G. Alginate as immobilization matrix for cells. Trends in Biotechnology 1990;8:7178 62. Augst AD, Kong HJ, Mooney DJ. Alginate hydrogels as biomaterials. Macromolecular Biosciences 2006;6:623633. 63. Normand V, Lootens DL, Amici E, Plucknett KP, Aymard P. New insight into agarose gel mechanical properties. Biomacromolecules 2000;1:730738. 64. Fatin Rouge N, Milon A, Buf fle J. Diffusion and partitioning of solutes in agarose hydrogels: the relative influence of electrostatic and specific interactions. Journal of Physical Chemistry 2003;107:1212612137.
204 65. Young T H, Chuang W Y, Hsieh M Y, Chen L -W, Hsu J -P. Assessment an d modeling of poly(vinyl alcohol) bioartificial pancreas in vivo Biomaterials 2002;23:34953501. 66. Isayeva IS, Kasibhatla BT, Rosenthal KS, Kennedy JP. Characterization and performance of membranes designed for macroencapsulation/implantation of pancrea tic islet cells. Biomaterials 2003;24:34833491. 67. Sakai S, Ono T, Ijima H, Kawakami K. Newly -developed aminopropyl -silicate immunoisolation membrane for a microcapsule -shaped bioartificial pancreas. Annals of the New York Academy of Sciences 2001;944(1) :277 283. 68. Massoud TF, Gambhir SS. Molecular imaging in living subjects: seeing fundamental biological processes in a new light. Genes & Developments 2003;17:545580. 69. Lu X, Dang H, Middleton B, Campbell Thompson M, Atkinson MA, Gambhir SS, Tian J, K aufman DL. Long -term monitoring of transplanted islets using positron emission tomography. Molecular Therapy 2006;14(6):851856. 70. deLonlay P, Simon -Carre A, Ribeiro M J, Boddaert N, Giurgea I, Laborde K, Bellane Chantelot C, Verkarre V, Polak M, Rahier J, Syrota A, Seidenwurm D, Nihoul -Fekete C, Robert J J, Brunelle F, Jaubert F. Congenital hyperinsulinism: pancreatic [18F]fluoro -L dihydroxyphenylalanine (DOPA) positron emission tomography and immunohistochemistry study of DOPA decarboxylase and insulin secretion. The Journal of Clinical Endocrinology & Metabolism 2006;91(3):933940. 71. Yang Y, Dubois A, Qin X-P, Li J, Haj AE, Wang RK. Investigation of optical coherence tomography as an imaging modality in tissue engineering. Physics in Medicine and Biol ogy 2006;51:16491659. 72. Gambhir SS. Molecular imaging of cancer with positron emission tomography. Nature Reviews 2002;2:683693. 73. Fowler MJ, Virostko J, Chen Z, Poffenberger G, Radhika A, Brissova M, Shiota M, Nicholson WE, Y. S, Hirshberg B, Harlan DM, D. JE, Powers AC. Assessment of pancreatic islet mass after islet transplantation using in vivo bioluminescence imaging. Transplantation 2005;79(7):768776. 74. Chan WC -W, Maxwell DJ, Gao X, Bailey RB, Han M, Nie S. Luminescent quantum dots for multip lexed biological detection and imaging. Current opinion in biotechnology 2002;13:4046. 75. Biju V, Itoh T, Anas A, Sujith A, Ishikawa M. Semiconductor quantum dots and metal nanoparticles: syntheses, optical properties, and biological applications. Anal y t ical and Bioanal y tical Chemistry 2008;391(7):24692495. 76. Tan W, Sendemir -Urkmez A, Fahrner LJ, Jamison R, Leckband D, Boppart SA. Structural and functional optical imaging of three -dimensional engineered tissue development. Tissue Engineering 2004;10(1112):17471756.
205 77. MacDonald JM, Grillo M, Schmidlin O, Tajiri DT, James TL. NMR spectroscopy and MRI investigation of a potential bioartificial liver. NMR in Biomedicine 1998;11:5566. 78. Schoeniger J, Aiken N, Hsu E, Blackband S. Relaxation time and di ffusion NMR microscopy of single neurons. Journal of Magnetic Resonance Series B 1994;103:261273. 79. Johnson KMR, Leavitt GD, Kayser HWM. Total body MR imaging in as little as 18 seconds. Radiology 1997;202:262267. 80. Armstrong P, Keevil SF. Magnetic r esonance imaging -1: Basic principles of image production. British Medical Journal 1991;303(6793):3540. 81. Keevil SF. Spatial localization in nuclear magnetic resonance spectroscopy. Physics in Medicine and Biology 2006;51:R579R636. 82. Kimmich R, Hoepf el D. Volume -selective multipulse spin echo spectroscopy. Journal of Magnetic Resonance 1987;72(2):379384. 83. Moonen CTW, von Kienlin M, van Zijl PCM, Cohen J, Gillen J, Daly P, Wolf G. Comparison of single -shot localization methods (STEAM and PRESS) for in vivo proton NMR spectroscopy. NMR in Biomedicine 1989;2(5/6):201208. 84. Haase A, Frahm J, Hanicke W, Matthaei D. 1H NMR chemical shift selective (CHESS) imaging. Physics in Medicine and Biology 1985;30(4):341344. 85. Hurd RE, Freeman DM. Metabolite specific proton magnetic resonance imaging. Proceedings of the National Academy of Sciences USA: Biochemistry 1989;86:44024406. 86. Bolan PJ, Meisamy S, Baker EH, Lin J, Emory T, Nelson M, Everson LI, Yee D, Garwood M. In vivo quantification of choline compounds in the breast with 1H MR spectroscopy. Magnetic Resonance in Medicine 2003;50(6):11341143. 87. Negendank W. Studies of human tumors by MRS: a review. NMR in Biomedicine 1992;5:303324. 88. Long RCJ, Papas KK, Sambanis A, Constantinidis I. In vitro monitoring of total choline levels in a bioartificial pancreas: 1H NMR spectroscopic studies of the effects of oxygen level. Journal of Magnetic Resonance 2000;146(1):4957. 89. Hoult DI, Busby SJW, Gadian DG, Radda GK, Richards RE, Seeley PJ. Observation of tissue metabolites using 31P nuclear magnetic resonance. Nature 1974;252:285287. 90. Moon RB, Richards JH. Determination of intracellular pH by 31P magnetic resonance. Journal of Biological Chemistry 1973;248:72767278.
206 91. Papas KK, R. C. Long J, Con stantinidis I, Sambanis A. Role of ATP and Pi in the mechanism of insulin secretion in the mouse insulinoma TC3 cell line. Biochemistry Journal 1997;326:807814. 92. Papas KK, Long RCJ, Constantinidis I, Sambanis A. Effects of short -term hypoxia on a tran sformed cell -based bioartificial pancreatic construct. Cell Transplantation 2000;9:415422. 93. Taylor J, Deutsch C. 19F -nuclear magnetic resonance: measurements of [O2] and pH in biological systems. Biophysical Journal 1988;53:227233. 94. McGovern KA, Sc hoeniger JS, Wehrle JP, Ng CE, Glickson JD. Gel -entrapment of perfluorocarbons: a fluorine 19 NMR spectroscopic method for monitoring oxygen concentration in cell perfusion system. Magnetic Resonance in Medicine 1993;29:196203. 95. Gross JD, Long RCJ, Con stantinidis I, Sambanis A. Monitoring of dissolved oxygen and cellular bioenergetics within a pancreatic substitute. Biotechnology and Bioengineering 2007;98(1):261270. 96. Nth U, Grhn P, Jork A, Zimmermann U, Haase A, Lutz J. 19F MRI in vivo determinat ion of the partial oxygen pressure in perfluorocarbonloaded alginate capsules implanted into the peritoneal cavity and different tissues. Magnetic Resonance In Medicine 1999;42:10391047. 97. Bloembergen N, Purcell EM, Pound RV. Relaxation effects in magn etic resonance absorption. Physical Reviews 1948;73(7):679712. 98. Hyde JS, Jesmanowicz A, Grist TM, Foncisz W, Kneeland JB. Quadrature detection surface coil. Magnetic Resonance in Medicine 1987;4:179184. 99. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magnetic Resonance in Medicine 1990;16:192225. 100. Arnder L, Shattuck MD, Black RD. Signal -to -noise ratio comparison between surface coils and implanted coils. Magnetic Resonance In Medicine 1996;35:727733. 101. Hoult DI, Tomanek B. Use of mutually inductive coupling in probe design. Concepts in Magnetic Resonance (Magnetic Resonance Engineering) 2002;15(4):262285. 102. Hardy WN, Whitehead LA. Split ring resonator for use in magnetic resonance from 2002000 MHz. Review of Scientific Instruments 1981;52(2):213216. 103. Vullo T, Zipagan RT, Pascone R, Whalen JP, Cahill PT. Experimental design and fabrication of birdcage resonators for magnetic resonance imaging. Magnetic Resonance in Medicine 1992;24(2):243252.
207 104. Ede lstein WA, Hardy CJ, Mueller OM. Electronic decoupling of surface -coil receivers for NMR imaging and spectroscopy. Journal of Magnetic Resonance 1986;67:156161. 105. Hyde JS, Rilling RJ, Jesmanowicz A. Passive decoupling of surface coils by pole insertion Journal of Magnetic Resonance 1990;89:485495. 106. Boskamp EB. Improved surface coil imaging in MR: Decoupling of the excitation and receiver coils. Radiology 1985;157(2):449452. 107. Schnall MD, V. HS, J. S. Leigh J, B. C. A new double tuned probe for concurrent 1H and 31P NMR. Journal of Magnetic Resonance 1985;65:122129. 108. Fitzsimmons JR, Brooker HR, Beck BL. A transformer -coupled double -resonant probe for NMR imaging and spectroscopy. Magnetic Resonance in Medicine 1987;5:471477. 109. Li Y, Log an TM, Edison AS, Webb A. Design of small volume HX and triple resonance probes for improved limits of detection in protein NMR experiments. Journal of Magnetic Resonance 2003;164:128135. 110. Sinha N, Grant CV, Rotondi KS, Feduik Rotondi L, Gierasch LM, Opella SJ. Peptides and the development of double and triple resonance solid state NMR of aligned samples. Journal of Peptide Research 2005;65:605620. 111. Martin RW, Paulson EK, Zilm KW. Design of a triple resonance magic angle sample spinning probe for high filed solid state nuclear magnetic resonance. Review of Scientific Instruments 2003;74(6):30453061. 112. Tomanek B, Volotovskyy V, Gruwel MLH, McKenzie E, King SB. Double -frequency birdcage volume coils for 4.7T and 7T. Concepts in Magnetic Resonanc e Part B 2005;26B(1):1622. 113. Fitzsimmons JR, Brooker HR, Beck BL. A comparison of double tuned surface coil. Magnetic Resonance in Medicine 1989;10:302309. 114. Leroy Willig A, Bittoun J, Kan S, Gonord P. A simplified double tuned 31P -1H circuit for N MR in vivo spectroscopic experiments. Review of Scientific Instruments 1990;61(2):799801. 115. Schnall MD, Subramanian VH, Leigh JSJ, Gyulai L, McLaughin A, Cance B. A technique for simulations 1H and 31P NMR at 2.2T in vivo Journal of Magnetic Resonance 1985;63:401405. 116. Chang L H, Chew WM, Weinstein PR, James TL. A balanced -matched double tuned probe for in vivo 1H and 31P NMR. Journal of Magnetic Resonance 1987;72:168172. 117. Kan S, Gonord P. Single input double -tuned mutually coupled surface coi l circuits: an analysis. Review of Scientific Instruments 1991;62(10):24272429.
208 118. Schnall MD. Probes tuned to multiple frequencies for in -vivo NMR. In: Rudin M, de Beer R, Cady EB, Garwood M, Link J, Moonen CTW, Morris PG, Mller S, van Ormondt D, Schnall MD, K. U, van Zijl PCM, editors. In vivo Magnetic Resonance Spectroscopy 1: Probeheads and radio frequency pulses, Spectrum Analysis. Volume 26, NMR Basic Principles and Progress. Berlin, Heidelberg: Springer -Verlag; 1992. p 3363. 119. Gonen O, Murphy -Boesch J, Li CW, Padavic -Shaller K, Negendank WG, Brown TR. Simultaneous 3D NMR spectroscopy of protondecoupled fluorine and phosphorus in human liver during 5 -fluorouracil chemotherapy. Magnetic Resonance in Medicine 1997;37(2):164169. 120. Eleff SM, S chnall MD, Ligetti L, Osbakken M, Subramanian VH, Chance B, Leigh JSJ. Concurrent measurements of cerebral blood flow, sodium, lactate, and high-energy phosphate metabolism using 19F, 23Na, 1H, and 31P nuclear magnetic resonance spectroscopy. Magnetic Resonance Imaging 1988;7:412424. 121. McNichols RJ, Wright SM, Wasser JS, Cot GL. An inductively coupled, doubly tuned resonator for in vivo nuclear magnetic resonance spectroscopy. Review of Scientific Instruments 1999;70(8):34543456. 122. Schmidt HC, Good ing CA, James TL, Gonzalez -Mendez R, James JL. Comparison of in vivo 31P NMR spectra of the brain, liver, and kidney of adult and infant animals. Pediatric Radiology 1986;16:144149. 123. Schmidt HC, Gooding CA, James TL. In vivo 31P -MR spectroscopy of the liver in the infant rabbit to study the effect of hypoxia on the phosphorus metabolites and intracellular pH. Investigative Radiology 1986;21:156161. 124. Farmer THR, Johnson GA, Cofer GP, Maronpot RR, Dixon D, Hedlund LW. Implanted coil MR microscopy of renal pathology. Magnetic Resonance In Medicine 1989;10:310323. 125. Ford JC, Shlansky-Goldberg RD, Golden M. MR microscopy of the arterial wall in an experimental model of atherosclerosis: preliminary results. Journal of Vascular and Interventional Radi ology 1997;8(1):9399. 126. Summers RM, Hedlund LW, Cofer GP, Gottman MB, Manibo JF, Johnson GA. MR microscopy of the rat carotid artery after balloon injury by using and implanted imaging coil. Magnetic Resonance In Medicine 1995;33:785789. 127. Arnder L Zhou X, G.P.Cofer, Hedlund LW, Johnson GA. Magnetic resonance microscopy of the rat carotid artery at 300 megahertz. Investigative Radiology 1994;29(9):822826. 128. Malloy CR, Cunningham CC, Radda GK. The metabolic state of the rat liver in vivo measure d by 31P -NMR spectroscopy. Biochimica et Biophysica Acta 1986;885:111.
209 129. MacDonald JM, Schmidlin O, James TL. In vivo monitoring of hepatic glutathione in anesthetized rats by 13C NMR. Magnetic Resonance In Medicine 2002;48:430439. 130. Quistorff B, E ngkagul A, Chance B. 31 P NMR in the study of liver metabolism in vivo. Pharmacology Biochemistry & Behavior 1983;18(Suppl. 1):241244. 131. Shine NR, Xuan JA, Koretsky AP, Weiner MW. Determination of renal molar concentrations of phosphorus -containing met abolites in vivo using 31P NMR. Magnetic Resonance In Medicine 1987;4:244251. 132. Kivelitz D, Wagner S, Schnorr J, Wetzler R, Busch M, Melzer A, Taupitz M, Hamm B. A vascular stent as an active component for locally enhanced magnetic resonance imaging. I nvestigative Radiology 2003;38(3):147152. 133. Morikawa S, Inubushi T, Kito K, Amano S. Long term observation of in vivo 31P NMR spectra in carbon tetrachloride -intoxicated rabbit liver using implanted wireless surface coil. NMR Biomedicine 1995;8(1):38. 134. Frazer RQ, Byron RT, Osborne PB, West KP. PMMA: an essential material in medicine and dentistry. Journal of Long Term Effects of Medical Implants 2005;15(6):629639. 135. Licari JJ, Hughes LA. Chemistry and Properties of Coatings Polyesters -Polyvin yls Polystyrenes -Acrylics -Diallylphthalates -Polyamides Phenolics -Polysulfides. In: Licari JJ, Hughes LA, editors. Handbook of Polymer Coatings for Electronics Chemistry, Technology and Applications (2nd Edition): William Andrew Publishing/Noyes; 1990. p 85102. 136. Reichlin J, Bormashenko E, Sheshnev A, Pogreb R, Shulzinger E, Katzir A. Investigation of water penetration in polystyrene by use of polymer -coated AgClBr fibers and development of new sensor intended for the FEWS spectroscopy of organic compo unds in water. Proceedings of SPIE 2000 4129:305313. 137. Mata A, Fleischman AJ, Roy S. Characterization of polydimethylsiloxane (PDMS) properties for biomedical micro/nanosystems. Biomedical Microdevices 2005;7(4):281293. 138. Kannan RY, Salacinski HJ, Butler PE, Hamilton G, Seifalian AM. Current status of prosthetic bypass grafts: a review. Journal of Biomedical Material Research Part B: Applied Biomaterials 2005;74(1):570581. 139. Hoult DI, Richards RE. The signal to -noise ratio of the nuclear magnetic resonance experiment. Journal of Magnetic Resonance 1976;24:7185. 140. Peterson DM, Beck BL, Duensing GR, Fitzsimmons JR. Common mode signal rejection methods for MRI: reduction of cable shield currents for high static magnetic field systems. Concepts in Magnetic Resonance Part B (Magnetic Resonance Engineering) 2003;19B(1):18.
210 141. Regan F, MacCraith BD, Walsh JE, O'Dwyer K, Vos JG, Meaney M. Novel Teflon coated optical fibres for TCE determination using FTIR spectroscopy. Vibrational Spectroscopy 1997;14(2):239246. 142. Terman FE. Radio engineers' handbook. New York and London: McGraw -Hill Book Company, Inc.; 1943. 143. Froncisz W, Jesmanowicz A, Kneeland JB, Hyde JS. Counter rotating current local coils for high -resolution magnetic resonance imaging. Magnetic Resonance in Medicine 1986;3:590603. 144. Jacobs G, Assefa A, Willig Onwuachi J. Q measurement and simulation for RF coils. 2008; Toronto, Ontario, Canada. Mira Digital Publishing. p 1082. 145. Chen C N, Sank VJ, Cohen SM, Hoult DI. The field dependence of NMR imaging: I. Laboratory assessment of signal to noise ratio and power deposition. Magnetic Resonance In Medicine 1985;3:722729. 146. Gabriel C, Gabriel S, Corthout E. The dielectric properties of biological tissues: I. Literature survey. Physics in Medicine and Biology 1996;41:22312249. 147. Gabriel C, Lau RW, Gabriel S. The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz t o 20 GHz. Physics in Medicine and Biology 1996;41:22512269. 148. Gabriel C, Lau RW, Gabriel S. The dielectric properties of biological tissues: III. Parametric models for the dielectric spectrum of tissues. Physics in Medicine and Biology 1996;41:22712293. 149. Shapiro L, Cohen S. Novel alginate sponges for cell culture and transplantation. Biomaterials 1997;18:583590. 150. Glicklis R, Shapiro L, Agbaria R, Merchuk JC. Hepatocyte behavior within three dimensional porous alginate scaffolds. Biotechnology and Bioengineering 2000;67(3):344353. 151. Simpson NE, Khokhlova N, Oca Cossio JA, McFarlane SS, Simpson CP, Constantinidis I. Effects of growth regulation on conditionally -transformed alginate insulin secreting cell lines in vitro. Biomaterials 2005;26:4 633 4641. 152. de Graaf RA, Brown PB, McIntyre S, Nixon TW, Behar KL, Rothman DL. High magnetic field water and metabolite proton T1 and T2 relaxation in rat brain in vivo Magnetic Resonance In Medicine 2006;56:386394. 153. Henkelman RM. Measurement of s ignal intensities in the presence of noise in MR images. Medical Physics 1985;12(2):232233.
211 154. Keltner JR, Carlson JW, Roos MS, Wong ST, Wong TL, Budinger TF. Electromagnetic fields of surface coil in vivo NMR at high frequencies. Magnetic Resonance In Medicine 1991;22(2):467480. 155. Nijm GM, Sahakian AV, Swiryn S, Carr JC, Sheehan JJ, Larson AC. Comparison of self gated cine MRI retrospective cardiac synchronization algorithms. Journal of Magnetic Resonance Imaging 2008;28:767772. 156. Hiba B, Richar d N, Janier M, Croisille P. Cardiac and respiratory double self -gated cine MRI in the mouse at 7T. Magnetic Resonance in Medicine 2006;55:506513. 157. Larson AC, White RD, Laub G, McVeight ER, Li D, Simonetti OP. Self -gated cardiac cine MRI. Magnetic Reso nance in Medicine 2004;51:93102. 158. Cassidy PJ, Schneider JE, Grieve SM, Lygate C, Neubauer S, Clarke K. Assessment of motion gating strategies for mouse magnetic resonance at high magnetic fields. Journal of Magnetic Resonance Imaging 2004;19:229237. 159. Griffin JL, Mann CJ, Scott J, Shoulders CC, Nicholson JK. Choline containing metabolites during cell transfection: an insight into magnetic resonance spectroscopy detectable changes. FEBS Letter 2001;509:263266. 160. Volland NA, Mareci TH, Simpson NE Development of an inductively-coupled MR coil system for imaging and spectroscopic analysis of an implantable bioartificial construct at 11.1T. Magnetic Resonance In Medicine 2009:(submitted). 161. Beck BL, Jenkins KA, Rocca JR, Fitzsimmons JR. Tissue -eq uivalent phantoms for high frequencies. Concepts in Magnetic Resonance Part B (Magnetic Resonance Engineering) 2004;20B(1):3033. 162. Harris FJ. On the use of Windows for Harmonic Analysis with the Discrete Fourier Transform. Proceedings of the IEEE 1978; 66(1):5183. 163. Mao J, Mareci TH, Andrew ER. Experimental Study of Optimal Selective 180 Radiofrequency Pulses. Journal of Magnetic Reso n ance 1988;79(1):110. 164. Tanuus A, Garwood M. Adiabatic pulses. NMR in Biomedicine 1997;10:423434. 165. Chew WM, Chang L H, Flamig DP, James TL. The SWIFT method for in vivo localized excitation (SMILE). Journal of Magnetic Resonance 1987;75:523528. 166. Gruetter R. Automatic localized in vivo adjustment of all -first and second -order shim coils. Magnetic Resonance I n Medicine 1993;29:804811. 167. Schnall MD, Subramanian VH, Leigh JS. The application of overcoupled tank circuits to NMR probe design. Journal of Magnetic Resonance 1986;67:129134. 168. Grover FW. Inductance calculations: Dover Publications; 1946. 286 p
212 169. Joseph PM, Fishman JE, Mukherji B, Sloviter HA. In vivo 19F NMR imaging of the cardiovascular system. Journal of Computer Assisted Tomography 1985;9(6):10121019. 170. Gross JD. Non -invasive monitoring of oxygen concentrations and metabolic functions in pancreatic substitutes. Atlanta, GA: Georgia Institute of Technology; 2007. 189 p. 171. Haake EM, Brown RW, Thompson MR, Venkatesan R. Introductory signal acquisition methods: Free induction decay, spin echoes, inversion recovery and spectroscopy. Mag netic resonance imaging. Volume 1. New York: John Wiley & Sons, Inc.; 1999. p 111137. 172. Bernstein MA, King KF, Zhou XJ. Basic Pulse sequences. Handbook of MRI pulse sequences. Volume 1. Boston: Elsevier Inc.; 2004. p 579647. 173. Shah P. Magnetic reso nance characterization of skeletal muscle adaptations after incomplete spinal cord injury. Gainesville, FL: University of Florida; 2008. 245 p. 174. Stabler CL, Wilks K, Sambanis A, Constantinidis I. The effects of alginate composition 1310. 175. Simpson NE, Stabler CL, Simpson CP, Sambanis A, Constantinidis I. The role of the CaCl2guluronic acid interaction on algi nate encapsulated TC3 cells. Biomaterials 2004;25:26032610. 176. Zekorn T, Horcher A, Siebers U, Schnettler R, Klck G, B. H, Zimmermann U, Bretzel RG, Federlin K. Barium -cross linked alginate beads: a simple, one -step method for successful immunoisolate d transplantation of islets of Langerhans. Acta Diabetologica 1992;29(2):99106. 177. Petruzzo P, Cappai P, Ruiu G, Brotzu G. Cell Microencapsulation: A new method. Transplantation Proceedings 1994;26:35073508. 178. Black SP, Constantinidis I, Cui H, Tuck er Burden C, Weber CJ, Safley SA. Immune responses to an encapsulated allogenic islet -cell line in diabetic NOD mice. Biochemical and Biophysical Research Communications 2006;340:236243. 179. Axel L, Summers RM, Kressel HY, Charles C. Respiratory effect s in two -dimensional Fourier transform MR imaging. Radiology 1986;160:795801. 180. Batchelor PG, Atkinson D, Irarrazaval P, Hill DLG, Hajnal J, Larkman D. Matrix description of general motion correction applied to multishot images. Magnetic Resonance in M edicine 2005;54:12731280. 181. Odille F, Vuissoz P -A, Marie P -Y, Felblinger J. Generalized reconstruction by inversion of coupled systems (GRICS) applied to free -breathing MRI. Magnetic Resonance in Medicine 2008;60:146157.
213 182. Beck BL, Letzen BS, Bashi rullah R, Mareci TH. Wireless control of an implantable coil system for MRI/S. In: 16th Scientific Meeting and Exhibition of the International Society for Magnetic Resonance in Med i cine Proceedings, editor; 2008; Toronto, Ontario, Canada. p 1111.
214 BIOGRAP HICAL SKETCH Nelly Aline Volland was born in Bourgoin Jallieu, France, in 1979. She was successively raised in Bourgoin Jallieu, Montceau, Faverges, Mercury, and Luxeuil -les -Bains, due to her parents numerous relocations She quickly decided to concentr ate on sciences since physics and mathematics were her favorite subjects. She received her scientific baccalaurat with honors from Lyce Jean Moulin in Albertville, France, where she graduated high school in June 1997. She then joined the pre -engineerin g school Cycle Prparatoire Polytechnique in Nancy (CPP, France) to gain the scientific skills necessary to enter the Ecole Nationale Suprieure de Physique de Grenoble (ENSPG, France) where she was accepted in September 1999. She chose to focus on physi cal instrumentation during the second year of engineering school to get closer to the area of specialization she was truly interested in, Biomedical Engineering. During her third and last year of engineering school she enrolled at the University of Flori da where she finally specialized in the biomedical engineering with a concentration in biomedical imaging and signal processing. This opportunity allowed her to be awarded a dual degree, one French and one American. She received her Diplme d'Ingnieur ( master equivalent) in physics from the Institut National Polytechnique de Grenoble (INPG, France) in September 2002 and was awarded a Master of Science in biomedical engineering from the University of Florida in December 2003. To further specialize in th e field of biomedical engineering she enrolled in the Ph.D program at the University of Florida in August 2003. Her primary focus has been on biomedical imaging instrumentation. She hope s to continue developing and improving imaging methods to see smaller and smaller biological entities and contribute to the understanding of diseases and the development of their treatment.