<%BANNER%>

Diffuse Optical Tomography

Permanent Link: http://ufdc.ufl.edu/UFE0044646/00001

Material Information

Title: Diffuse Optical Tomography Imaging Multiple Structural and Functional Features
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Jiang, Ruixin
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: tomography
Biomedical Engineering -- Dissertations, Academic -- UF
Genre: Biomedical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Diffuse Optical Tomography has drawn more and more interests in the biomedical field over the recent couple of decades due to its ability to noninvasively recover not only tissue structural information but also functional and molecular properties.  The contrasts that optical parameters could demonstrate in DOT are usually higher than those of the conventional methods. Based on these contrasts, different approaches had been developed applying DOT for imaging, and so far lots of efforts were spent on detecting breast cancer by imaging tissue absorption and scattering coefficients as well as hemoglobin concentration and oxygen saturation level. In this work, we tried to expand the ability of DOT in breast cancer detection by introducing Phase-contrast diffuse optical tomography (PCDOT). PCDOT uses near-infrared diffusing light to non-invasively reconstruct tissue refractive index (RI) distribution. RI depends on the tissue’s physical and chemical properties and previous study revealed that it might serve as a promising imaging parameter in breast cancer detection. We’ve first developed a2-step method to improve the PCDOT image both qualitatively and quantitatively at single-wavelength; then we’ve introduced a multispectral PCDOT algorithm to more efficiently reconstruct RI simultaneously with other tissue functional parameters and attempted to improve this algorithm by different structural regularization methods. Measuring hemodynamic changes, oxygen delivery and cerebral blood flow is important for locating and interpreting pathological variations associated with epileptic disorders. We then further expanded the application of DOT by presenting a method of dynamic, noninvasive andfunctional diffuse optical brain imaging that is conducted simultaneously with hippocampus CA1 local field potential recordings for anesthetized rats under resting conditions and during acute chemoconvulant provoked seizures. By illuminating the scalp with near-infrared light and recovering, the backward scattered light were collected and three-dimensional (3D) absolute tissue optical absorption images with high temporal resolution were obtained using a finite-element based reconstruction algorithm. The measured tissue absorption changes were validated with optic-intrinsic-signals measurement. In the focal seizure model, the seizure focus could be identified using the technique denoted by local variations of tissue absorption level as well as hemoglobin and cerebral blood flow changes.The findings are consistent with general observations in seizures of significant local cerebral metabolism increase. Successive absorption images along with EEG signals demonstrated linearity relationships from the neurovascular coupling study, suggesting cerebral metabolism closely matches demand from neuronal changes. This preclinical study suggests that this technique is feasible to be applied to human study and can provide insights into brain function and mechanisms of seizure disorders.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Ruixin Jiang.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Jiang, Huabei.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044646:00001

Permanent Link: http://ufdc.ufl.edu/UFE0044646/00001

Material Information

Title: Diffuse Optical Tomography Imaging Multiple Structural and Functional Features
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Jiang, Ruixin
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: tomography
Biomedical Engineering -- Dissertations, Academic -- UF
Genre: Biomedical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Diffuse Optical Tomography has drawn more and more interests in the biomedical field over the recent couple of decades due to its ability to noninvasively recover not only tissue structural information but also functional and molecular properties.  The contrasts that optical parameters could demonstrate in DOT are usually higher than those of the conventional methods. Based on these contrasts, different approaches had been developed applying DOT for imaging, and so far lots of efforts were spent on detecting breast cancer by imaging tissue absorption and scattering coefficients as well as hemoglobin concentration and oxygen saturation level. In this work, we tried to expand the ability of DOT in breast cancer detection by introducing Phase-contrast diffuse optical tomography (PCDOT). PCDOT uses near-infrared diffusing light to non-invasively reconstruct tissue refractive index (RI) distribution. RI depends on the tissue’s physical and chemical properties and previous study revealed that it might serve as a promising imaging parameter in breast cancer detection. We’ve first developed a2-step method to improve the PCDOT image both qualitatively and quantitatively at single-wavelength; then we’ve introduced a multispectral PCDOT algorithm to more efficiently reconstruct RI simultaneously with other tissue functional parameters and attempted to improve this algorithm by different structural regularization methods. Measuring hemodynamic changes, oxygen delivery and cerebral blood flow is important for locating and interpreting pathological variations associated with epileptic disorders. We then further expanded the application of DOT by presenting a method of dynamic, noninvasive andfunctional diffuse optical brain imaging that is conducted simultaneously with hippocampus CA1 local field potential recordings for anesthetized rats under resting conditions and during acute chemoconvulant provoked seizures. By illuminating the scalp with near-infrared light and recovering, the backward scattered light were collected and three-dimensional (3D) absolute tissue optical absorption images with high temporal resolution were obtained using a finite-element based reconstruction algorithm. The measured tissue absorption changes were validated with optic-intrinsic-signals measurement. In the focal seizure model, the seizure focus could be identified using the technique denoted by local variations of tissue absorption level as well as hemoglobin and cerebral blood flow changes.The findings are consistent with general observations in seizures of significant local cerebral metabolism increase. Successive absorption images along with EEG signals demonstrated linearity relationships from the neurovascular coupling study, suggesting cerebral metabolism closely matches demand from neuronal changes. This preclinical study suggests that this technique is feasible to be applied to human study and can provide insights into brain function and mechanisms of seizure disorders.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Ruixin Jiang.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Jiang, Huabei.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044646:00001


This item has the following downloads:


Full Text

PAGE 1

1 DIFFUSE OPTICAL TOM O GRAPHY: IMAGING MULTIPLE STRUCTURAL AND FUNCTIONAL FEATURES By RUIXIN JIANG 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 2012

PAGE 2

2 2012 Ruixin Jiang

PAGE 3

3 To my dear mom

PAGE 4

4 ACKNOWLEDGMENTS First, I would like thank Dr. Huabei Jiang, for his support and guidance during the past six years of my pursuit for the Ph. D. He brought me insights into the field of research and demonstrated to me the art of optical imaging. His enthusiasm and meticulous attitude towards research will continuously inspire me for my future career. Secondly, I would like to thank my academic committee members: Dr. Mingzho u Ding, Dr. Sihong Song, and Dr. Rosalind Sadleir, for their comments and suggestions on my research. Thirdly, I would like to thank Dr. Paul Carney, Professor from the Department of Pediatrics of University of Florida, for helping developing the animal study protocol. I would also like to thank my research colleagues: Dr. Zhen Yuan, for his assistance with the algorithm development; Dr. Qizhi Zhang, for his help with all kinds of hardware manip ulations; Dr. Junli Zhou, for his help of conducting all the animal surgeries of my research; Dr. Lei Yao and Dr. Xiaoping Liang, for their help with my research projects; Mr. Lijun Ji and Mr. Tao Zhang, for their help with my experiments. Their help is hi ghly appreciated. Fourthly, I would like to thank my parents for their continuous encouragement and unconditional love. And thanks to my a ntie, for helping me go through a hardest time of my life. At last, thanks to NIH and DoD for the financial support on my research projects.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 14 1 INTRODUCTION ................................ ................................ ................................ .... 16 1.1 History of Optical Imaging ................................ ................................ ................. 16 1. 2 Breast Cancer and Optical Imaging ................................ ................................ .. 17 1.3 Epilepsy and Optical Imaging ................................ ................................ ........... 19 1.4 Optical Diffusion Theory ................................ ................................ .................... 25 1.5 DOT Reconstruction Algorithm: Forward solution and Inverse Solution Procedures ................................ ................................ ................................ .......... 27 2 MULTISPECTRAL PHASE CONTRAST DOT IMAGING: PHANTOM EXPERIMENTAL STUDIES ................................ ................................ .................... 30 2.1 Me asurement of Refractive Index Distribution and Concentration .................... 30 2.1.1 Reconstruction Methods ................................ ................................ .......... 30 2.1.2 Development of Multispectral PCDOT Algorithm ................................ ..... 33 2.1.3 Numerical Simulations ................................ ................................ ............. 36 2.2 Imaging of Refractive Index Distribution and Concentration in Heterogeneous Tu rbid Media ................................ ................................ .............. 37 2.2.1 Phantom Study Methods ................................ ................................ ......... 37 2.2.2 Results and Conclusions ................................ ................................ ......... 38 3 PHASE CONTRAST DOT IMAGING: IN VIVO STUDY ................................ .......... 50 3.1 Methods ................................ ................................ ................................ ............ 50 3.2 Results ................................ ................................ ................................ .............. 51 3.2.1 Case studies: Invasive ductal carcinomas ................................ ............... 51 3.2.2 Case studies: Benign modules ................................ ................................ 52 3.3.3 Statistical Analysis ................................ ................................ ................... 53 3.3.4 Conclusions ................................ ................................ ............................. 54 4 FAST DOT IMAGING SYSTEM OF EPILEPSY AND ITS CALIBRATION .............. 63 4.1 DOT Imaging System ................................ ................................ ........................ 63 4.1.1 Light Unit (LED) ................................ ................................ ....................... 64

PAGE 6

6 4.1.2 Galvanometers ................................ ................................ ........................ 64 4.1.3 CCD Camera ................................ ................................ ........................... 65 4.1.4 System Timing ................................ ................................ ......................... 65 4.2 DOT Imaging System Operations ................................ ................................ ..... 65 5 FAST DOT IMAGING OF EPILEPSY: PHANTOM EXPERIMENTAL STUDIES .... 72 5.1 Measurement of Absorption Distribution in Heterogeneous Turbid Media ........ 72 5.1.1 Materials and Methods ................................ ................................ ............ 72 5.1.2 Numerical Simulations ................................ ................................ ............. 73 5.2 Imaging of Absorption Distribution in Heterogeneous Turbid Media ................. 73 5.2.1 Methods and Materials ................................ ................................ ............ 73 5.2.2 Results and Conclusions ................................ ................................ ......... 74 6 FAST DOT IMAGING OF EPILEPSY: IN VIVO STUDY ................................ ......... 77 6.1 Methods and Materials ................................ ................................ ...................... 77 6.1.1 Animal Preparation and Measurement Protocol ................................ ...... 77 6.1.2 Electrophysiological Recordings ................................ .............................. 78 6.1.3 Optical Recording of Intrinsic Signals ................................ ...................... 78 6.1.4 Neurovascular Coupling Data Analysis ................................ ................... 79 6.2 Absorption Distributions of Epileptic Animals ................................ .................... 81 6.2.1 Case Studies: Absorption Distribution ................................ ..................... 81 6.2.2 Validation of Absorption Images ................................ .............................. 82 6.2.3 Results and Discussion ................................ ................................ ........... 83 6.3 Coupling study ................................ ................................ ................................ .. 84 6.3.1 Case Stu dies: Linearity of Optical Signals and EEG ............................... 84 6.3.2 Validation of the Linearity Coupling ................................ ......................... 84 6.3.3 Discussions ................................ ................................ ............................. 85 7 MULTISPECTRAL DOT IMAGING OF EPILEPSY: IN VIVO STUDY ..................... 98 7.1 Methods and Materials ................................ ................................ ...................... 98 7.1.1 Multispectral DOT Imaging System ................................ ......................... 98 7.1.2 Animal Preparation and M easurement Protocol ................................ ...... 98 7.1.3 Multispectral DOT Data Analysis: Hemoglobin ................................ ........ 99 7.1.4 Multispectral DOT Data Analysis: Cerebral Blood Flow ......................... 100 7.2 Results and Discussions: Multiple parameters ................................ ............... 103 8 ULTRA FAST MULTISPECTRAL DOT IMAGING OF EPILEPSY: IN VIVO STUDY ................................ ................................ ................................ .................. 111 8.1 Methods and Materials ................................ ................................ .................... 111 8.2 Results and Discussions ................................ ................................ ................. 111 9 CONCLUSIONS AND FUTURE STUDIES ................................ ........................... 119 9.1 Conclusions ................................ ................................ ................................ .... 119

PAGE 7

7 9.2 Future Studies ................................ ................................ ................................ 120 LIST OF REFERENCES ................................ ................................ ............................. 122 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 133

PAGE 8

8 LIST OF TABLES Table page 2 1 Values of RI and the Glucose Concentration used in the phantom study ........... 39 2 2 Comparison of the actual and recovered values of RI of the target .................... 39 2 3 Concentration of absorption chromophores and RI in the background and the four targets used for numerical simulation case 1 ................................ .............. 40 2 4 Concentration of absorption chromophores and RI values in the background and the target used for numerical simulation case 2 ................................ .......... 40

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Geometry of split elements ................................ ................................ ................. 41 2 2 Case 1 simulation geometry ................................ ................................ ............... 42 2 3 Reconstructed images of RI (a), concentrations of Hbo2 (b), Hb (c), and water (d) for case 1. ................................ ................................ ............................ 43 2 4 Reconstructed images of RI (a), concentrations of Hbo2 (b), Hb (c), and water (d) for case 1 with 1% random noise. ................................ ....................... 44 2 5 Case 2 simulation geometry. ................................ ................................ .............. 45 2 6 Reconstructed images of RI (a, e, I, m), concentrations of Hbo2 (b, f, j, n), Hb (c, g, k, o), and water (d, h, l, p) for case 2 (a d). ................................ ................ 46 2 7 Phantom geometry. R1 =50mm, R2 =5mm, d =14mm. ................................ ......... 47 2 8 RI images reconstructed from phantom measurements without the two step method where the absorption and scattering coefficients were and for both the background and the target. The target had (a) 1% glucose concentration ( n =1.3312), (b) 2% glucose concentration ( n =1.3332), (c) 3% glucose concentration ( n =1.3353), and (d) 5% glucose concentration ( n =1.3393). ................................ ................................ ................... 48 2 9 RI images reconstructed from phantom measurements with the two step method. The absorption and scattering coefficients were and for both the background and the target. The target had (a) 1% glucose concentration ( n =1.3312), (b) 2% glucose concentration ( n =1.3332), (c) 3% glucose concentration ( n =1.3353), and (d) 5% glucose c oncentration ( n =1.3393). ................................ ................................ ................................ ......... 49 3 1 (a) Schematic of the 10 wavelenth, 64 by 64 source/detector channels DOT system; (b) Overall look of the DOT system; (c) ring structure that holds the source/detector fiber bundles. ................................ ................................ ............ 55 3 2 Reconstructed absorption (a), scattering ( b), and RI images with one step method (c) or two step method (d) for malignant case 1. Tumor is indicated by arrow. ................................ ................................ ................................ ............. 56 3 3 Reconstr ucted absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 2. Tumor is indicated by arrow. ................................ ................................ ................................ ............. 57

PAGE 10

10 3 4 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 3. Tumor is indicated by arrow. ................................ ................................ ................................ ............. 58 3 5 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 4. Tumor is indicated by arrow. ................................ ................................ ................................ ............. 59 3 6 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 5. Tumor is indicated by arro w. ................................ ................................ ................................ ............. 60 3 7 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for benign cas e 1. Lesion is indicated by arrow. ................................ ................................ ................................ ................. 61 3 8 Reconstructed absorption (a), scattering (b), and RI images with two step method (c) for b enign case 2. Lesion is indicated by arrow. ............................... 62 3 9 Reconstructed absorption (a), scattering (b), and RI images with two step method (c) for benign case 3. Lesion is indicated by arrow. ............................... 62 4 1 Schematic diagram of the experiment setup. While the CCD camera and galvanometers were controlled by one computer, the EEG recordings were controlled by ano ther computer. The two systems were correlated by the trigger signal from the CCD camera. ................................ ................................ .. 67 4 2 Molar extinction coefficient spect ra of hemoglobin. ................................ ............ 68 4 3 LED light bulbs applied in the study. ................................ ................................ ... 68 4 4 Galvanometers with 2 axis scanning heads. ................................ ...................... 69 4 5 Connection box for the 2 16bit daughter board. ................................ .................. 69 4 6 Self written LabView control panel for the system operation. ............................. 70 4 7 PVCAMTEST program inspects the ROI selected for imaging. .......................... 71 5 1 Reconstructed cross section absorption images for the numerical simulation. Target is put (a) 1mm, (b) 2mm, (c) 3mm and (d) 5mm under the surface, respectively. ................................ ................................ ................................ ........ 75 5 2 Phantom experiments: (a) Geometry of the phantom, target was located at 1 and 2 mm under the phantom surfaces respectively in the two cases; (b) Reconstructed DOT images of the single embedded target at 1mm depth in X Y (z=2mm) and Y Z (x=7mm) plane; (c) Reconstructed DOT images of the single embedded target at 2mm depth in X Y and Y Z plane. ............................ 76

PAGE 11

11 6 1 (a) Locations of BMI injection site (solid dot) and scalp EEG electrodes (open circles), (b) DOT image domain. ................................ ................................ ......... 87 6 2 Sample EEG signal piece where interictal spikes and high frequency SWD ictal onset could be observed. ................................ ................................ ............ 88 6 3 EEG reco rding of a 10 second window with continuous SWDs. ......................... 89 6 4 (a h) z=2mm cross section images from the reconstructed 3D images at time injection. Seizure focus indicated by arrows. (i) Three dimensional image at ................................ ................................ ............................. 90 6 5 (a e) z=2mm cross section images from the reconstructed 3D images at time j) x=10mm cross section images from the recons tructed 3D images at time .......... 91 6 6 Three dimensional ........... 91 6 7 z=2mm cross section images from the reconstructed 3D images at succe ssive different time points. Seizure focus is indicated by arrow. ................ 92 6 8 z=3mm cross section images from the reconstructed reference experiment at ............................ 93 6 9 OIS images for the validation expe riment. (a) Brain images during the resting period; (b) Selected OIS images during 5~10 minutes after the BMI injection, seizure focus indicated by arrows. ................................ ................................ ...... 94 6 10 System linearity study. (a) EEG signals from four trials of ictal periods; (b) Reconstructed tissue absorption value at the seizure focus; (c) predicted absorption coefficient generated by the input stimuli and HRF. .......................... 95 6 11 System linearity study. (a) EEG signals from two trials of ictal periods; (b) Reconstructed tissue absorption value at the seizur e focus; (c) predicted absorption coefficient generated by the input stimuli and HRF. .......................... 96 6 12 Validation for the system linearity using OIS measurement. (a) EEG signals from three trials of ictal periods; (b) Measured reflectance change at the seizure focus; (c) predicted reflection generated by the input stimuli and HRF. ................................ ................................ ................................ ........................... 97 7 1 Schematic of the whole head multispectral DOT imaging system. ................... 105 7 2 EEG electrodes and tube for BMI drug infusion. ................................ ............... 106

PAGE 12

12 7 3 Agar solution was poured and solidified to fulfill the space between the ................................ ................................ 106 7 4 EEG recording for one animal from 0 to 8 minutes after the drug injection. Discharges are generated and developed in to continuous SWDs. .................. 107 7 5 Selected absolute absorption coefficient cross section images at different time points 2mm under the injection site of the brain. ......................... 108 7 6 Selected Hb and HbO2 cross section images at same time points as in Fig. 7 4 2mm under the injection site of the brain. ................................ ................... 109 7 7 Fitted volume normalized CBF cross section images at same time points as in Fig. 7 4 2mm under the injection site of the brain. ................................ ........ 110 8 1 EEG recording for one animal from 0 to 8 minutes after the drug injection. ..... 114 8 2 Selected continuous absolute absorption coefficient cross section images at different time points 1mm under the injection site of the brain during the resting state. ................................ ................................ .................... 115 8 3 Selected continuous absolute absorption coefficient cross section images at different time points 1mm under the injection site of the brain after the drug injection. ................................ ................................ ............................. 115 8 4 (a) A 70s EEG signal piece that is corresponding to 10 00 imaging frames of DOT. Significant changes of the spike frequencies could be observed in this piece. (b) Power spectra of the EEG signal. ................................ ..................... 116 8 5 Curves of the averaged HbT value over the same 70s as that of Fig. 8 4. Blue line represents the 2mm area of the injection site at the injection plane; red line represents the 2mm area on the plane 3mm from the injection le vel; green line represents the area 10mm from the injection site. ........................... 117 8 6 Detailed comparison of EEG abnormalities and corresponding optical signal response. (a d) EEG signals, each represented a piece of 2s length; (e h) corresponding response of optical signals ................................ ........................ 11 8

PAGE 13

13 LIST OF ABBREVIATION S BOLD Blood Oxygen Level Dependent CBF Cerebral Blood Flow CT Computed Tomography EEG Electroencephalography FEM Finite Element Method F MRI functional Magnetic Resonance Imaging H B DeoxyHemoglobin H B O2 OxyHemoglobin DOT Diffuse Optical Tomography MRI Magnetic Resonance Imaging NIR Near I nfrared PCDOT Phase contrast Diffuse Optical Tomography PET Positron emission Tomography RI Refractive Indedx ROI Region of Interest SO2 Oxygen Saturation

PAGE 14

14 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 DIFFUSE OPTICAL TOM O GRAPHY: IMAGING MULTIPLE STRUCTURAL AND FUNCTIONAL FEATURES By Ruixin Jiang December 2012 Chair: Huabei Jiang Major: Biom edical Engineering Diffuse Optical Tomography has drawn more and more interests in the biomedical field over the recent couple of decades due to its ability to noninvasively recover not only tissue structural information but also functional and molecular properties. The contrasts that optical parameters could demonstrate in DOT are usually higher than those o f the conventional methods. Bas ed on these contrasts, different appr oaches had been developed applying DOT for imaging, and so far lots of efforts were spent on detecting breast cancer by imaging tissue absorption and scattering coefficients as well as hemoglobin concentration and oxygen saturation level. In this work, we tried to expand the ability of DOT in breast cancer detection by introducing Phase contrast d iffuse optical tomography (PCDOT) PCDOT uses near infrared diffusing light to non invasively reconstruct tissue refractive index (RI) study revealed that it might serve as a promising imaging parameter in breast cancer detection. W step method to imp rove the PCDOT image both qualitatively and quantitatively at single multispectral PCDOT algorithm to more efficiently reconstruct RI simultaneously with

PAGE 15

15 other tissue functional parameters and attempted to improve this a lgorithm by different structural regularization methods. Measuring hemodynamic changes, oxygen delivery and cerebral blood flow is important for locating and interpreting pathological variations associated with epileptic disorders. We then further expand e d the application of DOT by present ing a method of dynamic, noninvasive and functional diffuse optical brain imaging that is conducted simultaneously with hippocampus CA1 local field potential recordings for anesthetized rats under resting conditions and during acute chemoconvulant provoked seizures. By illuminating the scalp with near infrared light and recovering, the backward scattered light were collected and three dimensional (3D) absolute tissue optical absorption images with high temporal resolution were obtained using a finite element based reconstruction algorithm. The measured tissue absorption changes were validated with optic intrinsic signals measurement. In the focal seizure model, the seizure focus could be identified using the technique deno ted by local variations of tissue absorption level as well as hemoglobin and cerebral blood flow changes. The findings are consistent with general observations in seizures of significant local cerebral metabolism increase. Successive absorption images alon g with EEG signals demonstrated linearity relationships from the neurovascular coupling study, suggesting cerebral metabolism closely matches demand from neuronal changes. This preclinical study suggests that this technique is feasible to be applied to hum an study and can provide insights into brain function and mechanisms of seizure disorders.

PAGE 16

16 CHAPTER 1 INTRODUCTION 1.1 History of Optical Imaging Diffuse optical tomography (DOT) uses non ionizing radiation to non invasively reconstruct both structural and functional properties of tissue by applying tomographic measurements of near infrared (NIR) diffusive light along the tissue boundary. Op tical imaging was called transillumination or diaphan ography at early time applied for breast imaging, in which the white light shine on a breast directly and the shadow of the breast was captured [ 1 ]. Later on, people use the red light and near infrared l ight (NIR, 600~800nm wavelength) for optical imaging since photons at these wavelengths could penetrate deeper into soft tissue [ 2 ]. O ptical imaging made significant progress during the last two decades due to the advanced mathematical models of optical t ransportation (see section 1.4 for details of the models). Meanwhile, developments of modern computers provide an easy accessed platform where the heavy computation of the mathematical models could be conducted, and advanced CPU and recent application of GPU further accelerated the computation process. Also, more advanced optical measurement tools become available in recent decades, such as photodiodes, photomultiplier tube, and CCD cameras. They are able to trace even single photons, detect the amplitude and phase of light, and thus provide accurate measurements that can lead to better quality imaging. DOT, as a promising imaging technique, has been widely investigated in areas of breast cancer detection and functional brain imaging [ 3 7 ]. There are three major types of DOT systems, including time domain, frequency domain and continuous wave (CW) [ 8 11 ]. While time and frequency domain based systems may provide more optical

PAGE 17

17 information than CW measurements, CW has been favored by some researchers due to i ts low cost and hardware simplicity and has been proven to be feasible in both phantom and in vivo clinical data [ 5 11 15 ]. 1.2 Breast Cancer and Optical Imaging Breast cancer is the second most common type of cancer after lung cancer worldwide. Similar to other type of cancers, breast cancer is caused by the cancer cells that change and grow out of control, invading into normal tissue and spread throughout the body. Breast cancer begins in breast tissue, which is made up of lobules and the ducts that con nect the lobules to the nipple. When constrained within the lobules and ducts where breast cancer initiated, the cancer stage is called in situ. The majority of in situ breast cancer is ducal carcinoma in situ (DCIS), which accounted for ~83% of local brea st cancer diagnosed during 2004 2008, while lobule carcinoma in situ (LCIS) accounted for ~11% of female in situ breast cancer diagnosed [ 16 ]. The cancer cells may metastasize to other tissues of the body through lymphatics or blood as they develop. The se riousness of breast cancer is closely related to the stage of the disease. In clinical settings, the TNM system classifies tumors by tumor size and how far it spread within the breast and nearby organs (T), lymph node involvement (N), and the presence or a bsence of distant metastases (M). The earlier the cancer is detected, the less likely that the cancer cells would metastasize and the higher possibility that the cancer to be cured. According to the American Cancer Society, in 2011, an estimated 230,480 ne w cases of invasive breast cancer will be diagnosed among women and approximately 39,520 women are expected to die from breast cancer. In addition, about 2,140 cases of breast cancer are expected to occur among men [ 16 ].

PAGE 18

18 A tumor or lump in breast could be palpable when grows to centimeter size but the specificity of lesion detection using the physical examination is only ~ 20% to 30%. A s the predominant conventional approach for breast cancer detection, annual mammography screening had been introduced to pu blic. Despite its advantages, X ray mammography has low specificity and relatively lower sensitivity as the breast density increases [1 7 1 8 ] And mammography would put patients under risk with exposure to X ray radiations. T hus other conventional techniqu es are develop ed to improve the specificity and sensitivity of mammography, among which magnetic resonance imaging (MRI) showing the most significant improvement but remaining costly [1 9 ] and therefore not being suitable for routine breast screening. Posit ron emission tomography (PET) has good specificity in detecting breast cancer but has low spatial resolution and high cost as well. DOT is introduced as a noninvasive alternative for conventional breast detection methods which only depend on structural al terations. Unlike mammography where patients would be exposed to radiation, DOT has no side effects which make it possible to become a routine screening approach with its low cost. DOT is able to image both functional and morphological information of tissu e, including the concentration of oxy hemoglobin, deoxy hemoglobin, and water based on the absorption spectra; particle size and density distribution based on the scattering spectra. T he major imaging contrast of optical imaging of breast cancer resulted f rom the tumor related angiogenesis [ 20 ], which triggers increased optical absorption [ 21 2 2 ]. And metabolic imbalance of oxygen in tumors would also cause tissue hypoxia, leading to contrasts of oxygen saturation images [2 3 ]. The difference between abnormal and normal tissues

PAGE 19

19 could reach 100% in the NIR region due to the increased hemoglobin concentration [ 24 ].The blood volume in abnormal breast tissue can reach 400% compared to the normal tissue due to the increased blood v essels and dilations [ 25 ]. To date, there had been a lot of studies on imaging breast cancer using DOT and promising results had been reported [ 26 31 ]. W hile DOT shows high sensitivity towards breast cancer detection, it has a limited specificity [ 5 1 2 3 2 3 5] I n an effort to overcome this limitation we exploited phase contrast DOT ( PCDOT ) [ 3 6] by adding the possibility of reconstruction tissue refractive index (RI) due to the fact that phase contrast is generated by the spatial variation of tissue RI an d tissue RI depends on the tissue s physical and chemical properties. Y et phase contrast has been featured in optical microscopy [ 3 7, 3 8] phase contrast computed tomography (CT) [ 3 9, 4 0] ultrasound tomography [ 4 1] and optical coherent tomography (OCT) [ 4 2] From recent OCT measurements of RI in human ductal carcinoma simulated animal cancer models, significantly differentiated RI is demonstrated between tumor and normal tissues [ 4 3]. Our multispectral PCDOT algorithm combined the RI recovery with other i mportant functional parameters to further improve the quality of PCDOT and to attenuate the crosstalks between RI and absorption/scattering related coefficients, thus provided a more reliable criteria for the breast cancer detection. 1.3 Epilepsy and Optical Imaging Epilepsy is a common chronic neurological disease involving recurrent, unprovoked seizures that affects approximately 3% of the human population during their lifetime [44] S eizures are caused by the synchronous, rhythmic firing of a populat ion of neurons, lasting from seconds to minutes. A bout 60 70% of patients experienced focal or partial seizures while another 30 40% of patients have generalized seizures [45]

PAGE 20

20 E pilepsy in about 70% of patients could be controlled by medication [46] While for medically refractory seizures, resection of the epileptogenic zone may be considered and thus a series of presurgical evaluation will be taken to assess brain structure abnormality and clinical feature of seizures. F or patient with new onset seizures neuroimaging helps to determine whether the seizure is acute provoked or unprovoked and if any immediate treatment need to be taken [47] D etermination of whether there s an underlying brain lesion is one of the primary steps in evaluating new onset seiz ure disorders. C ommon causes of acute seizures are brain tumor, perinatal hypoxic or hypoxemic events and malformations of cortical development (MCDs) for young children, head trauma for young adults and stroke for elderly people [48] Potentially epilepto genic lesion detected by neuroimaging would affirm focal seizure disorders and differentiate them from primary generalized seizure disorders, which are based on either genetic or idiopathic rather than a focal epileptogenic lesion. T he distinction between these two types of seizure serves as the major standard for antiepileptic medication selection [49] Medically intractable epilepsy is defined as unsatisfactory seizure control despite optimized sequential use of 2 or more appropriate antiepileptic drugs [ 47] Epilepsy surgery is performed in patients with medically intractable epilepsy to improve seizure control and quality of life. A presurgical evaluation of patients with intractable epilepsy is applied to determine whether a patient has a single epilept ogenic focus and to localize the epileptogenic zone. The epileptogenic zone is the cortical region that is indispensable for the generation of seizures and that has to be removed to render a patient seizure free. The epileptogenic zone is a theoretical con struct, which is defined in terms of different cortical zones [50]

PAGE 21

21 T he seizure onset zone is the region in which the seizures actually originate. The symptomatogenic zone is the (sub)cortical region producing ictal symptoms. T he functional deficit zone is the part of cortex with an abnormal function in/between seizures, due to morphological or functional factors [51] Acquiring brain structural and functional information with imaging is essential in the diagnosis and management of epileptic seizure disorde rs. As the primary structure imaging approach, magnetic resonance imaging (MRI) detects surgically relevant lesions in up to 80% of patients who undergo temporal lobectomy [52] and in about 60% of those who under go frontal lobe surgery [53] MRI could loca te the lesion and surrounding structures accurately and identify structures known to serve critical brain functions. Another advantage of MRI over conventional computed tomography (CT) is that it allows 3D structural images. However, there are also many pa tients with active epilepsy in which structural imaging shows no abnormality. Thus advances have been made in functionally imaging the abnormal cerebral patho physiology associated with epileptic seizures. Major functional imaging modalities for epilepsy i nclude functional MRI (fMRI), positron emission tomography (PET) [54 57] ictal single photon emission computed tomography (SPECT) [58] and magnetic resonance spectroscopy (MRS). Along with hemoglobin changes, cerebral blood flow ( CBF ) and oxygen consumpt ion ( OC ) changes resulting from functional activations are all important components of the hemodynamic response in epileptic seizure disorders. Studying the functional information in vivo would provide insights in to the seizure generation and propagation while contributing to clinical seizure localization and understanding of complex metabolic and neuronal changes which these functional variations are inner

PAGE 22

22 related to. Knowledge of the co upling between neuronal activity and the associated hemodynamic response would also bring potential to predict seizure onset preceded its onset and thus benefit diagnosis and treatment of epilepsy [59] Therefore, there has been a highly increasing interes t in the subject of neurovascular coupling [60] On the other hand, malformations of cortical development (MCD) have been considered significantly as a major cause of seizures [61] which would demonstrate imaging contrast in cellular morphology imaging. Thus simultaneous imaging of the spatiotemporal characteristics of particle size/density, hemoglobin change, OC CBF and the cerebral metabolic rate of oxygen ( CMRO2 ) becomes important. As the primary functional image modality, fMRI detects the blood oxyg enation level depend (BOLD) signal that is associated with neuronal activities noninvasively [62] With the linear transform model, fMRI had been shown that the time course and amplitude of which could be predictable from the underlying neuronal responses and the fMRI variations could be colocalized with the underlying neuronal activity [63] However, the fact that signal from BOLD fMRI depends on mixtures of blood flow, blood volume and blood oxygenation instead of a specific physiologic feature makes it h ard to be quantified; the relatively low temporal resolution of fMRI would also disable s it from detecting important temporal information during the fast act brain activities in seizures; and moreover, ictal fMRI would not be routinely feasible clinically since the required cooperation level of fMRI scan might be hard to achieve for epileptic patients. Optical imaging based on the changes in light absorption of active neural tissue has been applied to exposed cortex to provide in vivo mapping of clinically relevant epileptiform events, such as interictal spikes, ictal onsets, ictal spread and secondary

PAGE 23

23 homotopic foci in animal models [64, 65] In the near infrared (NIR) region, NIR spectroscopy (NIRS) uses the ability of light in the near infrared region of 600 to 1000 nm wavelengths to penetrate tissue to depths of 8 10 cm, and could allow continuous and noninvasive monitoring of cerebral oxygen demand and supply by absorption spectra of oxyhemoglobin ( HbO2 ) and deoxyhemoglobin ( Hb ) in cerebral blood vessel s [66] In recent years there has been an increased interest in using diffuse optical tomography (DOT) for noninvasively imaging the human head and brain. DOT uses near infrared light that can pass through structures such as skull, penetrating the brain an d obtain images of the tissue optical properties and thus assess hemodynamic changes like regional blood volume and hemoglobin oxygen saturation. DOT can reconstruct 3D volumetric images of the head tomographically and hence can identify changes occurring in deeper tissues. Cerebral changes have been imaged by DOT in animal models during hypercapnia [67] focal ischemia [68, 69] and fore paw stimulation [70] Infant brain images during passive motor activation have been reported [71, 72] Recently, DOT was used to visualize the localized dynamics changes during seizure in rats that the seizure onset zone is localized by local dynamic changes of HbO2 Hb and total hemoglobin ( HbT ) concentrations during the ictal period [73] Compared to fMRI which is only s ensitive to changes in deoxyhemoglobin [74, 75] separate calculation of HbO2 and Hb could be done from DOT measurements, and the sum of the changes in the concentrations of these two species additionally provides measurement of total hemoglobin concentrat ion change, which is proportional to the cerebral blood volume change [76] Besides, DOT is able to reach a temporal resolution <100ms, which gives

PAGE 24

24 this technology advantage over fMRI [77] Though the spatial resolution of DOT is low (but still comparable to that of fMRI), DOT has the advantage of being inexpensive and portable, and therefore it is possible to conduct DOT scanning bedside. The ability of DOT to retrieve 3D tissue functional and molecular properties noninvasively distinguishes it from intrin sic optical signal imaging where incision is required for imaging [ 78 ] E pileptogenic foci induce a reduction of interictal glucose metabolism [ 79 ] while epileptic seizures increase cerebral metabolite dramatically coupled with cerebral vessels dilation [8 0 ] ; change of glucose concentration would cause tissue RI variation and the level of contrast is high enough to be sensitively detected by diffuse optical tomography (DOT) [ 8 1 ] C ompared to 18F fluorodeoxyglucose positron emission tomography (FDG PET), whi ch is the most commonly used technique for glucose metabolism imaging, DOT is inexpensive and portable, so that it could possibly be conducted bed side and used for continuous monitoring. Besides, the use of FDG PET is limited to the interictal period due to the short half of the tracer and its relatively long period of cerebral uptake (cerebral uptake of FDG took place over 40 minutes after the injection [8 2 ] ) whereas DOT won t have such limitation. In our previous work [73] we have reported in vivo 2D ima ges of HbO2 Hb and HbT in the rat brain during seizure onset using DOT. This pilot study suggested that DOT may be useful for epilepsy imaging due to its ability to dynamically localize epileptic foci and to map functional activities. However, this previ ous work was based on a DOT system with low temporal resolution (min per frame) and thus could not provide detailed interpretation of neuronal activity. In this current study, we present in vivo 3D images of focal seizure obtained from a newly developed fa st DOT system (700ms per

PAGE 25

25 frame) in reflection mode The new DOT system also allows concurrent electrophysiology (EEG) recordings during seizure onset. Moreover, the neurovascular coupling study conducted demonstrates linearity of the tissue absorption (mea sured by DOT) response to brain neuronal activities. The neurovascular coupling relationship observed from DOT was cross validated by optic intrinsic signal imaging. To our knowledge, this is the first study considering quantitative neurovascular coupling during the seizure ictal periods using DOT. 1.4 Optical Diffusion Theory Boltzmann transport equation describes incoherent photon propagation through highly scattering media. The time domain equation is described as following: (1 1) w here is the spatially varying RI distribution, is the radiance which is defined as the amount of energy perpendicular to the unit vector ( is measured i n unit ), is the absorption coefficient per unit length, is the scattering coefficient per unit length, is the scattering function which gives the probability that an energy packet travelling in direction travelling into direction is normalized according to is a source distribution per unit volume. However, the equation must be simplified to be mathematically manageable usually. By expanding the equation with spherical harmonics and truncate the ser ies at the term ( approximation), the quantities in Equation 1 1 could be expressed as: (1 2)

PAGE 26

26 (1 3) (1 4) where is the normalization factor, is the spherical harmonic of order L at degree m, is the Legendre polynomial of order L. approximation is obtained and the following equations are generated: (1 5) (1 6) where is the photon fluence, is the photon flux. The approximation can be further simplified by making the diffuse approximations that which is only justified when scattering coefficient is much larger than the absorption coefficient. Also assuming the photon source being isotropic yields Thus the transport equation could be app roximated to the following diffusion equations. In the time domain: (1 7) i n the frequency domain: (1 8) a nd in the continuous wave domain: (1 9)

PAGE 27

27 where is the diffusion coefficient, is the reduced scattering coefficient and is the isotropic source. For finite media, the boundary effects must also be taken into cons ideration. 1.5 DOT Reconstruction Algorithm: Forward solution and Inverse Solution Procedures The process of DOT image reconstruction includes the forward solution and the inverse solution. In the forward solution, light distribution in the medium is predi cted. Since it is impossible to acquire the analytical solution for the diffusion equation in the real case, finite element method (FEM) is applied to the study as it is able to solve the equation in heterogeneous medium with an arbitrary geometry. Accordi ng to previous studies [ 83 84 ] of FEM methods in continuous wave DOT, using the FEM discretization, coupling with type III boundary condition (1 10) the steady state diffusion equation could be transformed into ( 1 11) where is the boundary condition coefficient related to the internal reflection at the boundary; the elements of the matrix [ A ] are where the integrations are performed over the problem domain V ; is the diffusion coefficient where and are the absorption and reduced scattering coeffici ent ; and are locally spatially varying Lagrangian basis functions at node i and j respectively; is the source vector; where is the source strength

PAGE 28

28 and is the Dirac delta function for a source at ; M is the number of boundary nodes; is the photon density. In the forward solution procedure, the boundary condition coefficient the source strength and the initial value of absorption and diffusion coefficient should be determined by a preprocessing initial search opti mization scheme. and where and are measured and calculated photon density at i =1, M boundary locations. The squared difference of and are minimized as the function of the initial search parameters and the best initial values are generated based on the minimum squared difference. The inverse solution is based on the Taylor expansion or Newton method. Assuming the photon density are analytic functions of and D, and then can be Taylor expanded over an assumed (D, ) distribution, which is a perturbation away from some other distribution, : (1 12) w here and If the assumed optical property distribution is close to the true one, the high order items in the expansion can be negelected and we acquire (1 13) where

PAGE 29

29 (1 14) (1 15) (1 16) (1 17) and and are the reconstructed optical parameters. Regularization method is used to make Equation 1 13 invertible: (1 18) w here I is the identity matrix o f the size 2N x 2N, N is the node number of the finite element mesh, is the regularization parameter, J is the Jacobian matrix, and is the updating vector.

PAGE 30

30 CHAPTER 2 MULTISPECTRAL PHASE CONTRAST DOT IMAGING : PHANTOM EXPERIMENT AL STUDIES W e have previously reported an in vivo study of 35 breast masses (11 malignant cases and 24 benign cases) from 33 patients with phase contrast DOT ( PCDOT ) [ 8 5] T he results obtained from this study showed that malignant lesions generally had a decreased RI while benign lesions exhibited an increase d RI relative to the surrounding normal tissue. A sensitivity of 81.8% and a specificity of 70.8% were obtained from this study. While a significantly improved specificity was acquired by PCDOT compared to conventi onal DOT, the relatively noisy RI distribution recovered made it sometime difficult to identify the lesion, making the image inspection observer dependent. T he primary goal of th e current work is to develop a two step method for improv ing the quality of RI image reconstruction so that the image examination may be independent on the observer. In this two step method, a locally refined finite element mesh was created according to the reconstructed absorption/scattering image s using conventional DOT followed by incorporation of the structur al prior information into the iterative process for RI reconstruction We validate and evaluate this method using both phantom and in vivo data. 2.1 Measurement of Refractive Index Distribution and Concentration 2.1.1 Recons truction Methods A. Introduction to the Finite Element single wavelength PCDOT Reconstruction Algorithm O ur regularized nonlinear iterative reconstruction algorithm is based on the finite element solution to the following photon diffusion equation coupled wi th Type III boundary conditions [86, 87] :

PAGE 31

31 ( 2 1) (2 2 ) where is the photon density ; n is the refractive index; D is the diffusion coefficient ; is the absorption coeffici ent ; is the unit normal vector for the boundary surface; is a coefficient related to the internal reflection at the boundary and ; is the source strength and is the Dirac delta function for a sou rce at The diffusion coefficient can be written as where is the reduced scattering coefficient. Through a finite element discretization and other derived matrix relations by differentiation, a set of equations for inverse problem solution are obtained ( 2 3 ) ( 2 4 ) ( 2 5 ) in which the elements of the matrix [ A ] are where the integrations are performed over the problem domain V ; and are locally spatially varying Lagrangian basis fu nctions at node i and j respectively; is the source vector ; is the Jacobian matrix formed by / n at the boundary measurement sites ; a scalar and L the regularization matrix or filter matrix are used to reali ze the invertible system (Equation 2 5 ); is the update vector for RI,

PAGE 32

32 where N is the total node number of the finite element mesh used; and where and are measured and calculated photon density at i M boundary locations. To estimate the spatial distribution of n this quantity needs to be expanded in a similar manner to as a finite sum of unknown coefficient mu ltiplied by the locally defined Lagrangian basis function. The n distribution is updated iteratively through Equations 2 3 to 2 5 so that a weighted sum of the squared difference between measured and calculated photon density can be minimized. Diffusion and absorption coefficients are assumed constant during the RI reconstruction process. B. Adaptive Meshing To generate a locally re fined adaptive mesh, the absorption and scattering images are first reconstructed by conventional DOT where the lesion location/size can be confirmed in comparison with the x ray mammography or ultrasound (US). From these images, the maximum values ( max ) o f absorption/scattering coefficients in the lesion/target area and the minimum values ( min ) of absorption/scattering coefficients for the surroundings can be obtained. I f the values of both absorption and scattering coefficients at a nodal location are lar ger than where 0 < v < 1 then th is node is labeled as part of R egion I ; otherwise, the node is identified as part of Region II An element is split into four smaller elements ( Figure 2 1(a) ) if all three nodes associated with this element are part of Region I while an element is divided into two smaller ones ( Figure 2 1(b) ) if only two nodes associated with this element are part of R egion I C. Incorporation of Structur al a Prior i Informat ion

PAGE 33

33 T he regulari zation matrix L included in Eq uation 2 5 is commonly taken as the identity matrix and structur al prior information is iteratively incorporated in to the reconstruction process through the spatially varying regularization parameter [ 8 8] Here we use the Logarithm type regularization matrix which is constructed according to the priors obtained when generating the adaptive mesh. This regularization matrix can relax the smoothness constraints at the interface of different regions so that th e co variance of nodes within a region is basically realized [ 89 ] The elements of matrix L are: ( 2 6) where is the number of nodes within Region I and is the number of nodes within Region I I The Jacobian matrix is reassembled according to the region or tissue type it is associated with structural a priors 2.1.2 Development of Multispectral PCDOT Algorithm In the single wavelength PCDOT reconstruction algorithm, we made a first order approximation that both absorption and scattering coefficients are being constant during quantity of the reconstructed RI images through a two step method, it is still hard to retrieve the RI information at a lot of cases due to the highly heterogeneous distribution of absorption and scattering parameters and their big crosstalk on the RI image. Thus based on the general observation that the value of tissue RI does not have significant varies in the NIR region, we introduced more data from different wavelengths to the

PAGE 34

34 reconstruction to simultaneously recover RI along with other absorption and scattering derived chromophores. For breast tissue, the major abs orption chromophores are oxyhemoglobin (HbO2), deoxyhemoglobin (Hb), water and lipid [ 90 ]. Previous studies have demonstrated that hemoglobin levels in tumors tended to be larger while oxygen saturation levels are found to be lower than normal tissue [ 91 92 ]. Water content may also provide information for fiberadenoma or fibrocystic disease [ 93 ]. On the other hand, studies have shown that scattering spectra is correlated with tissue morphology [ 93 94 ] and in pathology that tumor cells are significantly enlarged compared to normal ones [ 95 ]. Tissue absorption is contributed by absorption chromophores with the concentration of for the th chromophore [ 96 ]. Thus the absorption spectra could be written as ( 2 7 ) w here indicates the absorption extinction coefficient of the th chromophore at wavelength The scattering spectra is correlated with particle size distribution and concentration under the Mie theory in the following relationship [ 96 97 ] ( 2 8 ) w here is the scattering efficiency, is the average cosine of scattering angles, is the particle size, is the RI of particles, is t he particle concentration/volume fraction, and is the particle size distribution. Assuming a Gaussian particle size distribution [ 96 ] ( 2 9 )

PAGE 35

35 w here is the average size of particles and is the standard deviation. Assume kinds of particles with different sizes and fix to be 1% as its impact is relatively small. The total scattering from all kinds of particles can be expressed by ( 2 10 ) w here is the volume fraction of the th kind of particles. Thus and as well as the RI could be reconstructed simultaneously with the following relationship ( 2 11 ) w here is the wavelength dependent photon density at boundary measurement sites. The Jacobian matrix can be obtained by the following calculations for ( 2 12 ) for ( 2 13 ) w hile has the same expression as that of the single wavelength reconstruction algorithm. Substituting Eq uation 2 12 into Equation 2 11 the system equation at all wavelengths c an be expressed as ( 2 14) w here is the number of wavelengths used. A Tikhonov regularization based iterative Newton method was used to solve the above equation after the Jacobian matrix was obtained.

PAGE 36

36 2.1.3 Numerical Simul ations For the following numerical simulation cases, the measured photon density used for reconstruction was calculated based on the finite element solution to Eq uation 2 1 The test geometry for case 1 is shown in Fig. 2 2 and the concentration of HbO2, Hb and water are listed in Table 2 3. Using measured data at five wavelengths (673, 733, 775, 840, and 922nm), the reconstructed concentration and RI images are shown in Fig. 2 3 without noise added in the data, and in Fig. 2 4 with 1% random noise added. These images implied that the absorption chromophore concentrations could be quantitatively recovered while RI could be qualitatively obtained both without and with noise. The crosstalk between the reconstructed parameters is not obvious without noise Add ing 1% random noise generated artifacts in the recovered images, especially for the water concentration image (Fig. 2 4 (d)), and the crosstalk is also enlarged in this image. However, the reconstructed RI, as well as concentration of HbO2 and Hb images are not deteriorated by the impact of the random noise. To see how much impact that different contrasts of RI and absorption chromophores have on each other, case 2 (a d) was performed. In these cases, we assumed only 1 target (Fig 2 5 ), but with different c ontrasts of RI and same contrasts of absorption chromophores. Values of RI and concentrations of Hb O 2, Hb and water are illustrated in Table 2 4. Using the same 5 wavelength as Case 1, the reconstructed images are shown in Fig. 2 6 where columns 1 4 repre sents recovered results for case 2 (a) (d), respectively. The reconstructed images indicated that the distribution of HbO2, Hb and water concentrations could generally be quantitatively recovered in all 4 cases and the RI image could be qualitatively obtai ned in case 2(a) and 2(b) when the RI has a negative contrast compared to the background. However, the reconstructed RI target

PAGE 37

37 has a small displacement in case 2(d) and in case 2(c) where the RI has a very low positive contrast (~1.5%). 2.2 Imaging of Refr active Index Distribution and Concentration in Heterogeneous Turbid Media 2.2.1 Phantom Study Methods Phantom experiments were conducted with a multispectral, multichannel diffuse optical tomography system, which was previously described in detail [ 98 ]. Fo r a 2D imaging experiment, light from a diode laser at 775 nm was transmitted sequentially to 16 source points at the phantom surface through an optical switch, and diffusing light was detected by 16 photodiodes. A set of 16x16 measured data was then input into the reconstruction algorithm to generate a 2D cross section image of the phantom. Four tissue like phantom experiments were conducted with different contrasts in RI between the target and the background. Tissue absorption ( ) and scattering ( ) were simulated for the background with India ink and Intralipid, respectively. Agar powder (2%) was used to solidify the mixed Intralipid India ink solution. Thus RI of the background was close to that of water ( n =1.33 at 775nm). One 10 mm diameter target was placed off center with various glucose concentrations to mimic different RI contrasts: ( 2 15 ) where [ C ] is the concentration of the glucose solution. Values of RI and their corresponding glucose concentrations used in this study are shown in Table 2 1. Geometry of the phantom is shown in Figure 2 7 A mesh of 717 nodes and 1368 triangular elements was appl ied in the reconstruction.

PAGE 38

38 2.2.2 Results and Conclusions Figures 2 8 and 2 9 respectively, present the reconstructed RI images for all four phantom cases without and with the two step method. We see that the i mages shown in Figure 2 8 are qualitatively go od in terms of target location and size, but have significantly overestimated RI values compared to the images shown in Figure 2 9 In addition, the images shown in Figure 2 9 also exhibit a better recovered target boundary relative to that shown in Fig. 2 8 To give a quantitative analysis of the two step method, we calculated the relative errors of the recovered target RI value for the four cases and listed the result s in Table 2 2. W e found that the relative errors range from 0.067% to 0.540%, which are s ignificantly improved compared to that from our previous study which used the hard priori regional reconstruction [ 8 1]

PAGE 39

39 Table 2 1 Values of RI and the Glucose Concentration used in the phantom study Glucose Concentration 1% 2% 3% 5% RI 1.3312 1.3332 1.3353 1.3393 Table 2 2 Comparison of the actual and recovered values of RI of the target RI Ideal 1.3312 1.3332 1.3353 1.3393 Calculated 1.3354 1.3260 1.3362 1.3437 Relative Error (%) 0.3 2 0.54 0.067 0.3 3

PAGE 40

40 Table 2 3. Concentration of absorption chromophores and RI in the background and the four targets used for numerical simulation case 1 Regions RI Concentration of absorption chromophores ( ) HbO2 Hb Water Background 1.33 40 40 30 Target 1 1.42 40 40 30 Target 2 1.33 80 40 30 Target 3 1.33 40 80 30 Target 4 1.33 40 40 60 Table 2 4. Concentration of absorption chromophores and RI values in the background and the target used for numerical simulation case 2 RI Concentration of absorption chromophores ( ) HbO2 Hb Water Background 1.33 40 40 30 Target in case 2(a) 1.27 80 80 60 Target in case 2(b) 1.30 80 80 60 Target in case 2(c) 1.35 80 80 60 Target in case 2(d) 1.42 80 80 60

PAGE 41

41 (a) (b) Figure 2 1 Geometry of split elements

PAGE 42

42 Figure 2 2 Case 1 simulation geometry

PAGE 43

43 Figure 2 3 Reconstructed images of RI (a), concentrations of Hbo2 (b), Hb (c), and water (d) for case 1. (a) (b) (c) (d)

PAGE 44

44 Figure 2 4 Reconstructed images of RI (a), concentrations of Hbo2 (b), Hb (c), and water (d) for case 1 with 1% random noise. (a) (b) (c) (d)

PAGE 45

45 Figure 2 5. Case 2 simulation geometry

PAGE 46

46 Figure 2 6 Reconstructed images of RI (a, e, I, m), concentrations of Hbo2 (b, f, j, n), Hb (c, g, k, o), and water (d, h, l, p) for case 2 (a d). (b) (f) (j) (n ) (c) (g) (k) (o) (d) (h) (l) (p) (a) (b) (c) (d)

PAGE 47

47 Figure 2 7. Phantom geometry. R1 =50mm, R2 =5mm, d =14mm.

PAGE 48

48 Figure 2 8. RI images reconstructed from phantom measurements without the two step method where the absorption and scattering coefficients were and for both the background and the target. The target had (a) 1% glucose concentration ( n =1.3312), (b) 2% glucose concentration ( n =1.3332), (c) 3% glucose concentration ( n =1.3353), and (d) 5% glucose concentration ( n =1.3393). (a) (b) (c) (d)

PAGE 49

49 Figure 2 9.RI images reconstructed from phantom measurements with the two step method. The absorption and scattering coefficients were and for both the background and the target. The tar get had (a) 1% glucose concentration ( n =1.3312), (b) 2% glucose concentration ( n =1.3332), (c) 3% glucose concentration ( n =1.3353), and (d) 5% glucose concentration ( n =1.3393). (a) (b) (c) (d)

PAGE 50

50 CHAPTER 3 PHASE CONTRAST DOT IMAGING : IN VIVO STUDY 3.1 Methods The clinical stu dy was approved by the institutional review board and was conducted in full compliance with the accepted standards for research involving human subjects. Signed informed consent from all study participants was obtained. In this study, 42 breasts from 42 di fferent patients (mean age 59; range from 32 82) were screened. Biopsy reports demonstrated 21 invasive carcinoma and 21 benign lesions. 29 patients were imaged by the same multi channel photodiodes based system used for the phantom experiments described e arlier [ 98 ], 9 patients were examined by a multi channel photomultiplier tubes (PMT) system [ 99 ] and 4 patients were screened by a single PMT based scanning system [ 100 ]. Data at one wavelength was used for the reconstruction, i.e., 775nm from the system d escribed in [ 98 ] and 785nm from those described in [ 99 100 ]. However, most of these patients were imaged by a 10 wavelenth, 64 by 64 source/detector channels DOT system [ 98 ] T he schematic of the system is shown in Fig 3 1(a) Generally, l ight beams from ten laser modules are transmitted to the optical switch, which sequentially passes one of the beams to 64 preselected points at the surface of the breast via source fiber bundles. The ring structure, which is detailed Fig 3 1( c ) holds the 64 source and 64 detection fiber bundles. Light from the bundles is sensed by the detection units, which convert the light intensity into voltage signals. The computer collects the signals though a data acquisition board. The DC motor near the ring is use d to adjust the diameter of the ring. Two CCD cameras are mounted underneath the ring to monitor the contact between the breast and fiber optics. The

PAGE 51

51 entire system is controlled by a LabVIEW program. During the examination, the patient would be positioned prone on the bed ( Fig 3 1( b ) ) and the breast imaged would be placed pendant through the opening. The bed could be moved vertically and the fiber optic array could be moved in and out to ensure that the rings are all in gentle contact with the breast. Al l lights in the room would be turned off as the imaging went on and it took about 30 minutes for a 10 wavelength scan. 3.2 Results 3.2.1 Case studies: Invasive ductal carcinomas Reconstructed images of absorption, scattering and RI from representative clin ical cases are shown in Figs. 3 2 ~3 6 The first case was a 66 year old woman who had an invasive ductal carcinoma that measured 1.9cm in maximal dimension. Both the absorption and scattering images exhibit marked increase in the region of tumor (indicate d by arrow in Figs. 3 2 a and 3 2 b), whereas the RI image without the two step method (Fig. 3 2 c) shows neither clear increase nor decrease in the tumor area corresponding to that of the absorption/scattering images. Fig. 3 2 d gives the reconstructed RI ima ge using the two step method where we see an identifiable decrease of RI at the tumor area (indicated by arrow). The second case was a 39 year old woman who had a palpable mass in the right breast (Fig 3 3 ). Biopsy showed infiltrating ductal carcinoma and mammography imaged the size of the mass ~2cm in diameter. Both absorption and scattering images exhibit increased value in the region of tumor corresponding to the mammography result. The 2 step method showed decreased RI at the tumor, which is consisten t with the one step method.

PAGE 52

52 The third case was a 50 year old woman with palpable abnormality in the right breast (Fig 3 4 ). Biopsy showed invasive ductal carcinoma with extensive ductal carcinoma in situ. Reconstructed absorption and scattering images showed marked increase at the tumor site. While the one step showed a relative increased RI, the two step method demonstrated decreased RI compared to the normal tissue. The fourth case was a 58 year old woman (Fig. 3 5 ). Biopsy showed the right breast had metastatic poorly differentiated carcinoma with ductal lobular features measuring 1.2cm. Absorption image showed a marked increased feature at the tumor site as indicated by arrow. RI recovered by the two step method exhibited decreased RI at the tumor si te. The fifth case was a 62 year old woman (Fig. 3 6 ). Biopsy showed infiltrating ductal carcinoma, moderately differentiated while mammography demonstrated a mass of 3cm at the biggest dimension. Both absorption and scattering images showed increased valu e at the abnormality compared to the background. The two step method demonstrated decreased RI at the same area, which is consistent with the one step method result. 3.2.2 Case studies: Benign modules The first case was a 60 year old woman with a 9mm diame ter benign nodule. Increased absorption and scattering coefficients are noticed in the lesion area (Fig. 3 7 a, 3 7 b), while the RI image without the two step method (Fig. 3 7 c) demonstrates moderate contrast in the lesion area. The RI image using the two s tep method (Fig. 3 7 d) presents marked increase in the lesion. The second case was a 22 year old woman with a 1.2cm diameter benign nodule (Fig 3 8 ). And the third case was a 46 year old with a 1.9cm lobulated, solid benign

PAGE 53

53 mass (Fig 3 9 ). Absorption and scattering images showed marked increase at the lesion, while the RI image with the two step method demonstrated marked increase at the same area. 3.3.3 Statistical Analysis We also calculated the sensitivity and specificity for cancer detection for the 4 2 cases examined using a prediction rule revealed in previous clinical studies [ 8 5]. In this prediction rule, a lesion will be diagnosed as malignant if its RI value is smaller than that of its surroundings and its absorption and scattering coefficients ar e larger than its surroundings; otherwise it is a benign lesion. For the possible physiological reason behind this prediction rule, the glucose metabolism in tumor have been proved to be significantly increased compared to normal tissue in animal models [ 1 01 ], and as discussed in Ref. 102 we suspect the glucose consumption is higher in tumor and lower in benign lesions. Using this rule, we found that the sensitivity (81%) is similar to that obtained previously [ 8 5], but the specificity is significantly imp roved (81%) over the previous study (71%). We consider the PCDOT method to be a potential approach to provide glucose metabolism information. In the current single wavelength PCDOT algorithm, we made a first order approximation that both absorption and sca ttering coefficients are being constant during the reconstruction process of RI. In future studies, we plan to adopt multi spectral reconstruction method to simultaneously recover RI along with other absorption derived chromophores such as oxy and deoxy h emoglobin concentrations, which allows the study of combining glucose metabolism with blood volume and oxygen metabolism for more complete diagnosis of breast cancer.

PAGE 54

54 3.3.4 Conclusions We have developed a two step PCDOT method which is able to improve the RI reconstruction qualitatively and quantitatively, making PCDOT a potentially observer independent approach for cancer diagnosis. This method has been confirmed by phantom experiments and 42 sets of clinical data. We expect to further evaluate this method using larger scale clinical data as well as applying this method for imaging other organs/diseases.

PAGE 55

55 Figure 3 1. (a) Schematic of the 10 wavelenth, 64 by 64 source/detector channels DOT system ; (b) Overall look of the DOT system; (c) ring structure that holds the source/detector fiber bundles. (a) (b) (c)

PAGE 56

56 Figure 3 2 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 1. Tumo r is indicated by arrow. (a) (b) (c) (d)

PAGE 57

57 Fig ure 3 3 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 2. Tumor is indicated by arrow. (a) (b) (c) (d)

PAGE 58

58 Figure 3 4 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 3. Tumor is indicated by arrow. (a) (b) (c) (d)

PAGE 59

59 Figure 3 5 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 4. Tumor is indicated by arrow. (a) (b) (c) (d)

PAGE 60

60 Figure 3 6 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for malignant case 5. Tumor is indicated by arrow. (a) (b) (c) (d)

PAGE 61

61 Figure 3 7 Reconstructed absorption (a), scattering (b), and RI images with one step method (c) or two step method (d) for benign case 1. Lesion is indicated by arrow. (a) (b) (c) (d)

PAGE 62

62 Figure 3 8. Reconstructed absorption (a), scattering (b), and RI images with two step method (c) for benign case 2. Lesion is indicated by arrow. Figure 3 9. Reconstructed absorption (a), scattering (b), and RI images with two step method (c) for benign case 3. Lesion is indicated by arrow. (a) (b) (c) (a) (b) (c)

PAGE 63

63 CHAPTER 4 FAST DOT IMAGING SYS TEM OF EPILEPSY AND ITS CALIBRATION We have gained experience in using DOT for b reast cancer detection and the results showed remarkably increased specificity compared to previous methods. The reason that DOT had been applied to breast cancer detection is mainly due to the fact that NIR light could penetrate soft tissue with low atten uations and thus optical properties could be retrieved. While DOT is able to re cover different optical properties that are directly related to tissue biological features, brain imaging of epilepsy drew our attention because of the large optical contrasts that epilepsy could produce induced by intense neuronal activities. Current imaging m odalities of brain are being either invasive or lack the ability of capturing fast dynamic information in the brain. However, the previous imaging system that we used for breast cancer detection could not be applied to the brain study due to its low tempo ral resolution. Imaging breast cancer is a relatively steady state process due to the s low evolvement of tissue angiogenesis, but brain imaging ought to be able to capture fast hemodynamic and molecular variations that are related to swift neuronal changes Thus, in order to study epilepsy, we built a new DOT system for three dimensional brain imaging that is able to r each a high temporal resolution. 4.1 DOT Imaging System For our experiment, we used a reflection mode based continuous wave (CW) DOT system (Fig. 4 1). In this system, light from a light emitting diode ( ~780nm, radiated power ~1000mW, SMB780 1100 01 I, Epitex Inc.) was controlled by an isolated pulse stimulat or (Model 2100, A M Systems) and delivered through 2 axis galvanometers (XLR8 Open Frame Head QS 7, Nutfield Tech.) to multiple source points consequently

PAGE 64

64 on the scanning surface. Two orthogonally placed linear polarizers were applied to eliminate specular reflections. The screening site was then imaged onto a 12 bit CCD camera (CoolSNAP EZ, Ph o tometrics) and the whole system was controlled by a self written LabView program. The individual components are detailed below. 4.1.1 Light Unit (LED) The fiber coupl ed LED unit at 780nm is used as CW light source. Compared to laser diode, LED is small and cheap, and the non coherent light from LED is less likely to harm eyes and skins. According to the spectra of molar extinction coefficient (Fig. 4 2), tissue absorpt ion towards hemoglobin and water would reach a relatively low level in the NIR region and thus light penetration can be enhanced. Each LED module is driven by a FPGA controller (CuteDigi Technologies, Inc.). The pigtail fiber is 100um in diameter and has a numerical aperture of 0.66. Figure 4 3 shows the NIR LED unit. The output power can be set from zero to maximum by adjusting the power supply control voltage from 0 to 10 volts. The lighting of the LED is triggered by a function generator. 4.1.2 Galvanome ters A set of programmable high performance open frame scan heads featuring galvanometers controlled by a LabView program is used to deliver light from LEDs to the scan surface (Fig 4 4 ). The scan angle of the heads could reach +/ 22 degrees in each of t he x/y direction and the small step could reach 270 usec with QD 3000 Servo Driver Board. The driver board is further connected to a 16bit daughter card and through digital interface cables and a connection box, receiving controls from the PC (Fig 4 5).

PAGE 65

65 4 .1.3 CCD Camera The 12 bit CCD camera is used as the detection unit in the experiment. The camera has a 1392x1040 imaging array with 6.45x6.45 um pixels. CoolSNAP LVDS cable connected the camera to the LVDS interface card on PC. TTL ouput is generated whil e exposing, providing temporal information of the data acquisition. 4.1.4 System Timing For a regular frame during the animal experiment with 5ms exposure time, a 32x32 image would take ~27ms to be captured, delivered and saved to the PC. The galvanometers operate at 1ms per switching action. Thus for 5x5 illumination position array scanning, the data collection time adds up to 700ms. 4.2 DOT Imaging System Operations Before the imaging experiment, the system should be powered for 15 minutes for warm up. Us ually, the LED light power supply and function generator are powered first, and then the galvanometers. The CCD camera should be turned on the last and turned off the first to ensure system stability. Since the region of interest (ROI) to be imaged is rela tively small compared to the whole FOV of the camera ( ~250 pixels in each direction is taken in the 1392x1040 array), a PVCAMTEST program is applied to ensure the ROI selected on the camera covers the cortex region that need to be imaged (Fig 4 7). In the animal experiments, 25 source points were evenly distributed on the flat region of the animal head extending approximately from 2mm anterior to 10mm posterior of the Bregma and 6mm to lateral of the midline. The whole field of view was divided into ~32x32 pixels on the CCD camera.

PAGE 66

66 Figure 4 6 shows the LabView control panel of the imaging system. Operator changes the exposure time, ROI on the camera, and scanning area of the scan heads. At last, the system runs automatically for the data acquisi tion. The collected data will then be calibrated and reconstructed using the FEM based algorithm for DOT imaging.

PAGE 67

67 Figure 4 1. Schematic diagram of the experiment setup. While the CCD camera and galvanometers were controlled by one computer, the EEG reco rdings were controlled by another computer. The two systems were correlated by the trigger signal from the CCD camera.

PAGE 68

68 Figure 4 2 Molar extinction coefficient spectra of hemoglobin. Figure 4 3 LED light bulbs applied in the study.

PAGE 69

69 Figure 4 4 Galvanometers with 2 axis scanning heads. Figure 4 5 Connection box for the 2 16bit daughter board.

PAGE 70

70 Figure 4 6 Self written LabView control panel for the system operation.

PAGE 71

71 Figure 4 7. PVCAMTEST program inspects the ROI selected for imaging.

PAGE 72

72 CHAPTER 5 FAST DOT IMAGING OF EPILEPSY: PHANTOM EX PERIMENTAL STUDIES 5.1 Measurement of Absorption Distribution in Heterogeneous Turbid Media 5.1.1 Materials and Methods A re gularized nonlinear iterative reconstruction algorithm was applied for image recovery and analysis which is based on the finite element solution to the following photon diffusion equation coupled with Type III boundary conditions: ( 5 1) ( 5 2) where is the photon density; D is the diffusion coefficient; is the absorption coefficient; n is the unit normal vector for the boundary surface; is a coefficient related to the internal reflection at the boundary and; is the source strength and is the Dirac delta function for a source at The diffusion coeffi cient can be written as where is the reduced scattering coefficient. In this algorithm, the 3D and distributions are updated iteratively through a set of derived equations for the inverse problem so that a weighted sum of the squared difference between measured and calculated photon density can be minimized. A time series based calibration method for DOT was used to reduce systematic errors and enhance image reco nstruction quality [103] The average measurement data during the resting state was used as the reference medium to generate and divided by the mean measurement after the seizure drug injection, forming the calibration matrix

PAGE 73

73 which would then be applied to the data at a specific time point after the drug injection for image reconstruction. 5.1.2 Numerical Simulations To validate the three dimensional FEM based reconstruction methods, several sets of numerical simulations were c onducted. Simulated data were generated using the forward solution model with absorption and scattering distribution and values given. A cylindrical target, 3 mm in diameter with 2mm length in the z direction, is placed 1mm, 2mm, 3mm, and 5mm under the sur face to simulate variations at different depths of the brain. The optical values are set as =0.007/0.028 and =1.0/2.0, for the background and target respectively. The forward data is then reconstructed with a 242x16 mesh with 25 source points and 242 detectors. Results from the simulations are displayed in Figure 5 1. From these Y Z plane cross sections, we can see that target at different depth were accurately recovered in the absorption images. It is valida ted for the reflection based three dimensional system where source and detectors are on the same plane, target with depth variations could be recovered by the FEM method. 5.2 Imaging of Absorption Distribution in Heterogeneous Turbid Media 5.2.1 Methods an d Materials Two tissue like phantom experiments were conducted to validate the DOT system. Tissue absorption ( =0.01 =0.04 ) and scattering ( =1.0 =4.0 ) were simulated for the background and target respectively with india ink and intralipid; 2% agar powder was used to solidify the mixed intralipid ink solution. One 3mm diameter, 2mm height target was placed with various depths in each of the

PAGE 74

74 background p hantoms. The geometry of the phantoms is shown in Fig. 5 2(a). The FOV of the camera was 12mm x 21mm and the FOV was divided into 24 x 44 detectors. 5.2.2 Results and Conclusions We first tested the ability of the DOT system imaging targets at different de pth with tissue mimic phantoms. With 25 source points, there were 26400 source detector pairs applied to the image reconstruction and the total reconstruction volume was 12mm x 21mm x 20mm. Fig. 5 2(b, c) shows that the embedded target could be reconstruct ed at the right location both on the X Y plane and the Y Z plane. Comparing the two Y Z plane images, the difference of the reconstructed target depth position is significant, which shows that the DOT system can reasonably recover 3D volumetric objects.

PAGE 75

75 Figure 5 1 Reconstructed cross section absorption images for the numerical simulation. Target is put (a) 1mm, (b) 2mm, (c) 3mm and (d) 5mm under the surface, respectively. (a) (b) (c) (d)

PAGE 76

76 Figure 5 2. Phantom experiments: (a) Geometry of the phantom, target was located at 1 and 2 mm under the phantom surfaces respectively in the two cases; (b) Reconstructed DOT images of the single embedded target at 1mm depth in X Y (z=2mm) and Y Z (x=7mm) plane; (c) Reconstructed DOT images of the single embedded target at 2mm depth in X Y and Y Z plane. (a) (b) (c)

PAGE 77

77 CHAPTER 6 FAST DOT IMAGING OF EPILEPSY: IN VIVO ST UDY 6.1 Methods and Materials 6.1.1 Animal Preparation and Measurement Protocol Six adult male Sprague Dawley rats (4 rats for the DOT scanning, 2 rats for reference and validation), weighting 220 to 280g, were included in this study. All procedures were conducted according to protocols approved by the University of Florida Institutional Animal Care and Use Commi ttee. During surgical procedures, animals were placed in a stereotaxic frame and anaesthetized with isoflurane (4%) and maintained with one third of the initial dose supplemented with 0.4 L/min oxygen. Body temperature was kept constant using a heating pad for each animal and the heart rate was continuously monitored. Generally, two incisions were made and three screw electrodes were implanted, two at 3mm anterior of the Bregma for electrophysiological (EEG) recording and another one posterior of the Lambda for ground reference. Another hole was then drilled on the skull over the left frontal lobe (with the same small piece of bone put back and covered the hole after the surgery) for later seizure drug injection (Fig. 6 1 ). Following the above surgical proce dures, isoflurane was stopped and 1.2g/kg urethane was injected intraperitoneally instead. We allow at least 30 minutes for the anesthetic stability before the experiments started. To induce focal seizures, local injection of Bicuculline (BMI, 10l, 1.9mM) by a mini syringe via the prepared hole was conducted. DOT measurements made before the BMI injection were used as calibration data and scans were conducted continuously for up to an hour after the BMI injection, along with whole course EEG monitoring.

PAGE 78

78 6. 1.2 Electrophysiological Recordings Two channels of EEG data were collected simultaneously with DOT using a 16 channel digital amplifier system (R25 Bioamp processor, Tucker Davis Tech.). 2 mm screw electrodes were applied to record the neuronal activities In addition, two optical electrical converters were connected to one channel on the amplifier system to deliver CCD shutter trigger signals to the EEG system so that optical signals captured by DOT could be synchronized with recorded electrical signals. EEG data were processed using self written MATLAB programs. Low pass filtered at a cut off frequency at 50Hz, we calculated the integrated signal response during each frame of the DOT scan and applied it to off line neurovascular coupling study. 6.1.3 Opti cal Recording of Intrinsic Signals To validate the DOT images and neurovascular coupling study results, we applied intrinsic optical signal (OIS) measurement to the same seizure stimulation protocol. There have been several studies using OIS to investigate brain activities including neocort ical seizures [64, 65] that this method had been able to image the seizure focus OIS measurement, with the same positions of EEG electrodes, the scalp of the animal was cut open and the skull was removed to expose the brain. 2% agar solution was poured and solidified on the brain to create a flat surface and eliminate specular reflections. Light from the LED source was directed and illuminated the whole brain and CCD was focused on the brain surface. Images of a 12x12mm area of the cortex were acquired every 33ms with 180x150 pixel resolution. For the data analysis, we consider the reflectance change from the mean pixel intensity as the inverse of the tissue absorptio n in order to qualitatively evaluate the neurovascular relationships.

PAGE 79

79 6.1.4 Neurovascular Coupling Data Analysis Numerous models had been applied to analyze the neural respo nse in imaging by lots of groups. In fMRI, Statistical parametric mapping (SPM) had been widely applied [104] The simple idea behind SPM is that images proceeded with voxel values that are, under the null hypothesis, distributed according to a known probability density function. The SPM method is usually test ed for activation, or regres sion on some explanatory variable. As the basis for SPM, general linear model (GLM) acts as the foundations for the variant of most analysis methods [105] for instance, the simple t tests on scans assigned to one condition to another, correlation coeffici ents between observed responses and boxcar stimulus functions in fMRI, or the evoked responses estimated using linear time invariant models. Bayesian inference would be applied where the null hypothesis based SPM method is rejected when the probability of obtaining the statistic is sufficiently small. Bayesian inference based upon the posterior distribution of the activation given the data, which could be computed from the probability that the activation exceeds some threshold. Dynamic models include convol ution models and temporal basis function where the hemodynamic response function is applied, which is the cornerstone for making statistical inferences about activations in fMRI with the GLM; the hemodynamic response, induced by any given trail type, can b e expressed as the linear combination of several basis functions of per stimulus time. Biophysical models as modified convolution models are input state output systems that start with a causal dynamic model of how re s ponses are generated and construct a ge neral linear observation model that allows estimating and inferring things about the parameters of that model, where convolution models design matrices are not informed by a forward model of how data are caused.

PAGE 80

80 To date there had been several studies on th e neurovascular coupling in epilepsy. In PET and fMRI, most studies have been performed on interictal rather than ictal events, which may elicit such a brief focal increase in deoxygenated hemoglobin as to be undetectable without higher strength magnets. S ome ictal fMRI studies have been done on generalized spike and wave events, which may not elicit an increase in HbR in the cortex, while some have been done without concurrent electrical recordings, rendering the timing of the imaging unclear with respect to the seizure onset. In optical spectroscopy, along with oxygen sensitive electrodes, there has been studies of the the onset of epileptic event [65] But the intrinsic optical spectroscopy is incapable of resolving subtle variations in the electrophysiology arising after seizure onset. Hemodynamic response function (HRF) models have been used to model the dynamic BOLD signals in fMRI study [106, 107] that a lin ear and time invariant system is assumed. The basis of the analysis is the comparison between the actual fMRI signal and a model of the expected respons e given the knowledge of generalized spike and wave timing. Statistical maps were obtained, indicating a t each voxel the level of correlation between the BOLD signal and the model, and among the results from the different models, the one yielding the highest t stat value was selected. The HRF is the impulse response function that the hemodynamic signal could be modeled as the convolution of HRF and input stimulus (neural activities): ( 6 1 ) where is the measured hemodynamic signal, and is the measured neural activity and is the HRF.

PAGE 81

81 In our neurovascular coupling study, in order to test the coupling between the measured tissue absorption changes and the neural activities, similar linear regression model was applied to predict the hemodynamic var iations. Since the measured EEG signal (i.e., the input stimulus), has a much higher sampling rate than the recovered optical signals, EEG signal was integrated in each of the event related period. Here we used the integrated power of the neural impulse re sponse as it shows robust linear relationship between it and the BOLD signal even if neurovascular coupling is nonlinear due to a purely vascular refractory effect [108] 6.2 Absorption Distributions of Epileptic Animals 6.2.1 Case Studies : Absorption Dist ribution For the DOT animal seizure experiments, 17 seizures were imaged in four rats. Following the BMI injection, EEG recordings show that interictal spikes were generated and seizures occurred periodically for up to 1h. Fig. 6 2 shows a 100s EEG recording window for one animal at 200s after the drug injection. And Fig. 6 3 shows a typical 10s continuous spike and wave discharge piece during the seizure onset. In each animal, the first electrographic seizure onset occurred at 2 3 m in following the BMI injection denoted by high frequency continuous spike and wave discharges (SWDs). Different amplitude of electrical signals could be observed between the two recording channels, which indicated the localized seizure onset. For the first case, Absolute absorption coefficient images at different time points were recovered. Fig. 6 4 (a) shows selected reconstruction absorption coefficient distribution of the brain at the depth of 2mm. Fig. 6 4 (b) illustrates the 3D image at the 140s time point. Seizure focus (indicated by arrows) with increased

PAGE 82

82 absorption coefficient could be seen in the reconstructed in vivo images despite the existence of boundary artifacts. For the second case, absorption images at different tim e points were reconstructed. Fig.6 5 6 6 point. Seizure focus (indicated by arrows) with increased abso rption coefficient could be seen in the reconstructed in vivo images. For the third case, the drug is injected at the primary somatosensory cortex and Fig. 6 7 shows selected reconstruction absorption coefficient distribution of the brain at the depth of 2 mm. Seizure focus (indicated by arrows) could be observed with marked increasing of absorption compared to its surrounding regions. 6.2.2 Validation of Absorption Images To confirm that changes of tissue absorption at the seizure focus were caused by epile ptic activities, reference/control experiment was conducted where same volume of saline was injected instead of BMI. Fig. 6 8 shows selected time points images after the bo undary artifact. To further validate the accuracy of the DOT images, Fig. 6 9 shows the OIS images from the reference experiment of the same stimulation protocol. As indicated by the OIS images, the position of the increased tissue absorption focus was co nsistent with the drug injection site and the size of the focus is very comparable with the DOT reconstruction images over time.

PAGE 83

83 6.2.3 Results and Discussion BMI is a well established seizure model that had been used in many aspects and the EEG recordings from our study had been very similar to those from previous studies [109, 110] Continuous high frequency SWDs from the EEG suggested seizure onset and it is also obvious that the closer the recording electrode to the BMI injection site, the more significa nt change of the EEG signal could be observed. For this localized acute seizure onset model, the seizure foci could be detected in all cases with a marked increase of local tissue absorption. In the shown case (Fig. 6 4 ), there is a clear maximum around ( x=5.5, y=6, z=2mm) relative to its surrounding basically throughout the 30 minute monitoring period after the drug injection. This is consistent with the general observation that epileptic seizures increase cerebral metabolism dramatically coupled with ce rebral vessels dilations, and localized cerebral blood volume increase should be observed in localized seizure onset, which would result in local increase of tissue absorption levels. Both the size and absorption value of this focus could be observed to ch ange significantly over time compared to the stable where the BMI solution was injected. Ictal cerebral blood flow had been suggested to increase around 150% 200% in PET study [111] the scale of which is comparable to what we have observed in the absorption images ( ~0.4 at the focus and ~0.2 at the surrounding).

PAGE 84

84 6.3 Coupling study 6.3.1 Case Studies : Linearity of Optical Signals and EEG To quantitatively study the recovered 3D absorption images and to test the hypothesis of the system linearity, neurovascular coupling study were conducted using the linear regression model. In the cu rrent study, we only investigated the ictal periods that are denoted by fast SWDs. HRF was generated based on a single ictal period and generally, the HRF has an initial dip phase of 2s and delay of undershoot phase of 10s. Then the same response function was applied to all other event related periods to predict the tissue absorption variations. The measured absorption value was calculated by the average of the seizure focus which usually measured ~2mm in diameter. Fig. 6 10 shows the measured and predicted results in several trials from one animal. The duration of the ictal periods varied from 15s 30s. Fig. 6 11 shows the measured and predicted results from a second animal. The duration of the ictal periods varied from 25s 50s. As indicated by the figures, the absorption variations could be qualitatively predicted by the input stimulus and the HRF that the predicted curve correlate with the actual change. Calculated correlation coefficient is ~0.20 for the given cases. This result would strongly suggest the linearity of the system and the linear regression model could be nicely applied to the study. 6.3.2 Validation of the Linearity Coupling OIS was used to validate the DOT reconstructed results qualitatively. As indicated by the reflectance images over time after the BMI injection, both the size and the value of the focus varied over time after the BMI injection and the size of the increased absorption area was comparable to what had been recovered from DOT. Similar kind of

PAGE 85

85 neurovascular coupling study had also been conducted and results are shown in Fig. 6 12 Here the change of reflectance could also be predicted using the same linear regres sion model that was used in the DOT study, which suggested consistent linearity of the system. 6.3. 3 Discussions So far there had been a number of studies, both on human and animals, on the quantitative coupling of hemodynamic and neural responses but had different conclusions. Local field potential from forepaw stimulation was observed to be linear with integration of BOLD and CBF generated from BOLD in animal subjects [119]; LFP from visual cortical response was found linear with amplitude of BOLD signal [120]; CBF from laser Doppler and optical intrinsic signal was found to be linear with summed neural activities in animal subjects [121]; in human subjects, somatosensory evoked potential was observed to be linear with BOLD signal amplitude [122]; deoxy h emoglobin from diffuse optical imaging was also found to be linear with MEG peak amplitudes in human subjects [123]. Meanwhile, nonlinear cases had been found also. BOLD from visual stimulus was observed to be nonlinear with MEG in human subjects [124] and some studies claimed hemoglobin percentage change from spectroscopic optical measurements to be linear with LFP from somatosensory cortex stimulus from animal subjects [125]. In these studies, the quantification of neural and hemodynamic signals varied. Peak amplitude, steady state amplitude, peak to peak difference, or time integral was used in different models. Lacking standard way of defining and quantifying neural and hemo dynamic signals before analysis may lead to disagreement of linearity of the models.

PAGE 86

86 In this study, we used the absolute value of reconstructed optical absorption coefficient as the hemodynamic response signal, and time integral of the EEG response as the neural response signal. Tissue optical absorption is directly related to hemoglobin concentration and tissue oxygen saturation, and EEG as the measurement of integrated synaptic activities, indicated the major utilization of cerebral energy. We found that tissue optical absorption is linearly related to EEG activities during the highly activated seizure onset period. This relationship implies close coupling between tissue oxygen metabolism and demand of neuronal energy in the intensive and comprehensive neu ronal activities. This understanding of the coupling between hemodynamic responses and neuronal activities would also assist the interpretation of perfusion based imaging techniques such as fMRI and PET.

PAGE 87

87 Figure 6 1. (a) Locations of BMI injection site (solid dot) and scalp EEG electrodes (open circles), (b) DOT image domain. (a) (b)

PAGE 88

88 Figure 6 2. Sample EEG signal piece where interictal spikes and high frequency SWD ictal onset could be observed.

PAGE 89

89 Figure 6 3 EEG recording of a 10 second window with continuous SWDs.

PAGE 90

90 Figure 6 4 (a h ) z=2mm cross section images from the reconstructed 3D images at BMI injection. Seizure focus indicated by arrows. ( i ) Three dimensional image (a) (b) (c) (d) (e) (f) (g) (h) (i)

PAGE 91

91 Figure 6 5 (a e) z=2mm cross section images from the reconstructed 3D images at (f j) x=10mm cross section images from the reconstructed 3D images at time Figure 6 6 Three (f) (g) (h) (i) (j) x y z (a) (b) (c) (d) (e)

PAGE 92

92 Figure 6 7 z=2mm cross section images from the reconstructed 3D images at successive different time points Seizure focus is indicated by arrow.

PAGE 93

93 Figure 6 8 z=3mm cross section images from the reconstructed reference experiment at time (a) (b) (c) (d)

PAGE 94

94 Figure 6 9 OIS images for the validation experiment. (a) Brain images during the resting period; (b) Selected OIS images during 5~10 minutes after the BMI injection, seizure focus indicated by arrows. (a) (b)

PAGE 95

95 Figure 6 10 System linearity study. (a) EEG signals from four trials of ictal periods; (b) Reconstructed tissue absorption value at the seizure focus; (c) predicted absorption coefficient generated by the input stimuli and HRF. (a) (b) (c)

PAGE 96

96 Figure 6 11 System linearity study. (a) EEG signals from two trials of ictal periods; (b) Reconstructed tissue absorption value at the seizure focus; (c) predicted absorption coefficient generated by the input stimuli and HRF. (a) (b) (c)

PAGE 97

97 Figure 6 12 Validation for the system linearity using OIS measurement. (a) EEG signals from three trials of ictal periods; (b) Measured reflectance change at the seizure focus; (c) predicted reflection generated by the input stimuli and HRF. (a) (b) (c)

PAGE 98

98 CHAPTER 7 MULTISPECTRAL DOT IM AGING OF EPILEPSY: I N VIVO S TUDY 7.1 Methods and Materials 7.1.1 Multispectral DOT Imaging System To further increase the temporal resolution of DOT imaging and to acquire functional information associated with seizure activities, a multispectral fast DOT imaging system was built. Th e details of the system are described elsewhere [112] Generally, instead of the previous reflection mode based imaging, the new system images the whole head. Two wavelengths of LED lights were chosen as the light source at 780nm and 850nm. Each wavelength contains 48 LEDs that are able to switch fast between one and another (50ns). FPGA is also used to swiftly and accurately control the LEDs. Two types of detectors, high sensitive APD (avalanche photodiode) and low sensitive SPD (silicon photodiode), are c ombined in the system to ensure a larger dynamic range. The highest sampling rate could reach 0.14 Hz after 100 times average of each detection time point, which is applied to reduce the system noises. All the source and detection fiber bundles are evenly placed on a ring holder within which the 7 1. Phantom studies have proved that the system has a better capability in imaging deeper targets compared to the reflection mode based sy stem. 7.1.2 Animal Preparation and Measurement Protocol Four adult male Sprague Dawley rats, weighting 220 to 280g, were included in this study. All procedures were conducted according to protocols approved by the University of Florida Institutional Animal Care and Use Committee. During surgical procedures, animals were placed in a stereota xic frame and anaesthetized with isoflurane (4%) and

PAGE 99

99 maintained with one third of the initial dose supplemented with 0.4 L/min oxygen. Body temperature was kept constant using a heating pad for each animal and the heart rate was continuously monitored. Dur ing the surgery, two incisions were made. One was made posterior of Lambda, and one screw electrode was implanted for ground reference. Another incision was made from 5mm anterior of the Bregma to 5mm posterior. As the EEG electrodes, two fine copper wires were implanted into the head 1mm left or right from the Bregma respectively. A small hole ~1mm in diameter was drilled 1mm posterior of the right electrode and a small tube was inserted into the hole to provide duct of later BMI drug introduction (Fig. 7 2). Following the above surgical procedures, isoflurane was stopped and 1.2g/kg urethane was injected intraperitoneally instead. We allow at least 30 minutes for the anesthetic stability before the experiments started. To fill the space between the holder ring and the animal head, liquid phantom above 40 degrees were infused into the space that it would cooled and solidified (Fig 7 3). Bicuculline (BMI, 10l, 1.9mM) injected into the brain by previously placed tube was used to induce focal seizures. DOT me asurements made before the BMI injection were used as calibration data and scans were conducted continuously for up to an hour after the BMI injection, along with whole course EEG monitoring. 7.1.3 Multispectral DOT Data Analysis : Hemoglobin For brain, the major absorption chromophores are also oxyhemogobin, deoxyhemoglobin and water as in the breast tissue. Thus the absorption spectra could be written as ( 7 1 )

PAGE 100

100 w here indicates the absorption extinction coefficient of the th chromophore at wavelength and is the concentration for the th chromophore. Referring to the molar extinction coefficient spectra, the composition of water at 780nm and 850nm to the abso rption is significantly low compared to those of Hb and HbO2. Thus the absorption spectra could be written as ( 7 2 ) given the above two wavelengths, the concentrations of Hb and HbO2 could be implied by solving Equation 7 1 ( 7 3 ) ( 7 4 ) Tissue absorptions and were reconstructed independently under each wavelength using the FEM reconstruction method; and were selected according to [ 113 ]. Thus oxyhemoglobin and deoxyhemoglobin maps could be generated based on the absorption distributi on from the two wavelengths. 7.1.4 Multispectral DOT Data Analysis : Cerebral Blood Flow So far different image modalities had been applied to study CBF, including PET, fMRI, laser Doppler, diffuse correlation spectroscopy, and laser speckle contrast imagi ng. However, among these image approaches, PET has limited spatiotemporal resolution [114] ; fMRI is based on the BOLD signal thus requires careful selection of the scaling factor towards deoxyhemoglobin concentration [115] ; laser Doppler flowmetry has limite d penetration depth and is only able to measure limited points [116] ; diffuse correlation spectroscopy may not be practical for continuous monitoring in humans although it had implied promising results [117] ; laser speckle contrast imaging is able to

PAGE 101

101 recover relative CBF with high spatiotemporal resolution but is not noninvasive. Thus there remains a demand for continuous and noninvasive imaging method of CBF with low cost and high spatiotemporal resolution. DOT as a noninvasive, economical, portable imaging method had been proved to be able to recover continuous tissue absorption, scattering, hemoglobin concentration and oxygen saturation with a high spatiotemporal resolution in the previous studies. Yuan [ 118 ] developed the application of DOT by establishing a new mathematical model that connects changes in CBF to changes of hemoglobin and oxygen saturation( SO 2 ), and thus the model is able to indirectly measure neuron induced vascular parameters including CBF. To model oxygen transport in a blood vessel, consider a one dimensional cylindrical vessel with R i and R o as the inner and outer radii, respectively, surrounded by other biological tissues. In addition, it is assumed that all the oxygen (O 2 ) diffusing out the segment is consumed in a tissue region. The law of mass conservation stipulates that the amount of O 2 lost from a vascular segment must be equal to the diffuse O 2 flux to the tissues, determined by the perivascular oxygen gradients. For a steady case, ( 7 5 ) in which (mL.s 1 ) is the volumetric BF into the ith segment, the volumetric BF out the segment, d i is the diameter of the ith segment, l i the length of the ith segment, HbT the total hemoglobin concentration in the blood (moles), the hemoglobin oxygen saturation flowing in the segment, oxygen saturation flowing out of the segment, J i the oxygen flux across t he vessel wall (moles O 2 cm 2 .s 1 ) and C b is the

PAGE 102

102 oxygen binding capability of hemoglobin ( C b =1.39mLO 2 /gmHb; C b =1. if the concentration of O 2 dissolved in plasma is considered). Through a series of deductions [], it could be concluded that, (7 6) where V tissue is the tissue volume and is assumed constant here, SO 2 is the oxygen saturation of tissue, Q is the mean BF for all the blood vessels inside the tissues and is specified as the mean BF of tissues, [ HbT ] blood i s the mean total blood hemoglobin concentration in the blood circulating through the tissues, OC is the mean oxygen consumption for the whole tissue volume V tissue and is the averaged hemoglobin oxygen saturation at the inlet(artery) and outlet(vena)of the tissues, respectively. This equation 7.6 is the developed mathematical model that connects changes in CBF known hemoglobin concentrations and oxygen saturations captured by DOT. Mean CBF can be recovered by fitting Equation 7 6 to time resolved tissu es oxygenation measurements. Equation 7 6 is an ordinary partial differential equation that can be solved iteratively by Runge Kutta 4 th order method coupled with finite element method. With a ny given initial values for OC and BF within the specified range, the fitting method is to optimize the OC and BF parameters based on the solution of the minimal squared difference between measured and calculated oxygen saturation values from multiple disc rete time points. To reconstruct BF and OC the initial parameters are given by: HbT blood =0.72mM, f =0.2, SO 2,ti =0.98.

PAGE 103

103 7.2 Results and Discussions: Multiple parameters For the focus animal seizure experiments, multiple seizure activities were imaged in the four rats. Fig. 7 4 shows the EEG recording for one animal 8 minutes after the drug injection, and it could be observed that interictal spikes were generated and devel oped into periodic seizures. Using the collected data, tissue absorption, oxyhemoglobin, deoxyhemoglobin and CBF were recovered by the previously described methods. It is noted that since the images were averaged 100 times at each detection point in order to reduce system noise, the interval between each adjacent frames is 7s. Figure 7 5 shows selected absolute absorption coefficient cross section images at different time points 2mm under the injection site of the brain both under 780nm and 850nm wavelengths. Seizure focus (indicated by arrows) with increased absorption coefficient could be seen in the reconstructed in vivo images. Based on the reconstructed absorption images, hemoglobin concentration distribution image s were generated, which is illustrated in Figure 7 6. The increasing dynamics in blood volume and oxygenation in the scalp and brain caused the variations of DOT measured dynamics during seizures and the marked increase of hemoglobin concentration at the s eizure focus is consistent with the general observations. Then mean CBF is fitted for each linear segment based on the Hb and HbO2 images. For each segment, 2 discrete hemoglobin and oxygen saturation measurement points are used for fitting. Figure 7 7 pre sented the volume normalized CBF images over the 5 time points in Fig. 7 5& 7 6. Increasing of CBF could be observed at the focus and the peak value of CBF (10 38ml/100ml/min) are in good agreement with reported CBF of rats (between 10 120 ml/100ml/min) [ 1 26 ].

PAGE 104

104 Simultaneous measurement of hemoglobin concentration, oxygen saturation, cerebral blood flow and EEG represented neuronal activities allowed a comprehensive assessment of quantified functional hemodynamic activities. In this study, we observed an incr under multiple trials. The results are consistent with the general observations of increased blood flow and oxygen consumption during the seizure activities and validated the multi spectral imaging system and algorithms for reconstructing multiple optical parameters.

PAGE 105

105 Figure 7 1 Schematic of the whole head multispectral DOT imaging system.

PAGE 106

106 Figure 7 2 EEG electrodes and tube for BMI drug infusion. Figure 7 3 Agar solution was poured and solidified to fulfill the space between the

PAGE 107

107 Figure 7 4 EEG recording for one animal from 0 to 8 minutes after the drug injection Discharges are generated and developed in to continuous SWDs.

PAGE 108

108 Figure 7 5. S elected absolute absorption coefficient cross secti on images at different time points 2mm under the injection site of the brain

PAGE 109

1 09 Figure 7 6. Selected Hb and HbO2 cross section images at same time points as in Fig 7 4 2mm under the injection site of the brain.

PAGE 110

110 Figure 7 7. Fitted volume normalized CBF cross section images at same time points as in Fig 7 4 2mm under the injection site of the brain.

PAGE 111

111 CHAPTER 8 ULTRA FAST MULTISPECTRAL D OT IMAGING OF EPILEP SY: IN VIVO STUDY 8 .1 Methods and Materials The multispectral DOT sys tem described in chapter 6 is able to image multiple optical parameters. However, averaging imaging 100 times to reduce the system noise greatly limited the temporal resolution of the system. Thus, by switching the data acquisition mechanism of the system and upgraded the data collection circuit, the temporal resolution of the system can reach its maximum performance at 14 Hz. The high sampling rate is crucial for the analysis of epilepsy especially the coupling study and could help to remove the noises fro m breathing and heart beating as well, thus providing better image qualities 8 .2 Results and Discussions With the same BMI induced focus seizure model, three animal experiments were conducted. Figure 8 1 shows the recorded EEG signal from one animal exte nded to 8 minutes after the drug injection. This animal had intensive seizures that several continuous spike and wave discharges pieces could be observed from the EEG recording. Figure 8 2 demonstrated 5 continuous cross section 780nm absorption images (1m m under the drug injection site) during the resting state. It is obvious that no focus like targets could be observed from the images. This is consistent with the EEG observation that no intensive neuronal activities were induced when the animal is under a nesthetization and resting. Figure 8 3 shows 5 continuous cross section 780nm absorption images (1mm under the drug injection site) after the BMI drug injection. Marked increasing of

PAGE 112

112 absorption could be seen at the seizure focus (indicated by arrow), which is markedly different from those of the resting state images. Figure 8 4 shows a 70s EEG signal piece that is corresponding to 1000 imaging frames of DOT. The average absorption of the seizure focus at two wavelengths is calculated. Total hemoglobin value s were reconstructed based on the absorption distributions (Fig 8 5 ). Dramatic change of HbT within 2 00 3 00ms had been observed in the curves, which could not be seen in previous relatively slow image systems. It is noted by comparing the power spectra of EEG and the HbT curves that when spikes were observed, significant change of the HbT could be found. In Figure 8 5, the blue line (focus) represented the averaged HbT value in a r=2mm area around the drug inje ction point at the seizure focus level, while the red line (surround ing ) represented averaged HbT value in a r=2mm area 3mm from the seizure focus plane. When spikes occur, hemoglobin at the focus and the surround ing would both drop to the same level and t hen increase. Hemoglobin at the focus increased much faster to 1.5 times more than the surrounding. The green line in Figure 8 5 indicated the average HbT value in an area >10mm distant from the focus, and no neural activity related change could be observe d. A detailed study of the EEG signal and optical response showed that the transient drop of HbT happened simultaneously with the neural abnormalities, and no significant advance or lag could be observed in the response (Figure 8 6). However, it should be noted that the EEG electrode placed is ~1mm distant from the drug injection site. Our data indicate s that, in a model of acute focus rat seizures, interitcal spikes are accompanied by transient hypometabolism surrounding the seizure focus followed by

PAGE 113

113 clea r increase of metabolism in the focus In previous studies, the epileptic surround around inhibition induced by interitctal spikes was considered to prevent interictal to ictal transitions [127]. This phenomenon has been observed in several studies [128, 129] but yet it was not well studied during in vivo seizures. In this study, we show clear evidence of interictal surround inhibition. The reason that focus area studied also went through a transisent inhibition is probably due to the spatial resolution of DOT itself that the responses from the surrounding area could not be completely isolated from the focus. However, the extent of hypermetabolism followed the intial suppression differentiated the seizure focus from the surrounding. Previous studies hypot hesized the etiology of the surround inhibition to be a passive steal phenomenon or from active shunting of blood initiated by vasoconstriction of vessels in the surrounding [130, 131]. Schwartz [132] reported preictal vasoconstriction followed by vasodila tion in a recent study, which is consistent with our observations. It was also reported in his study that vessels start from ~1.5mm distance from the injection site would go through preictal vasoconstrictions, which partially explained the drop of HbT in t he focus area.

PAGE 114

114 Figure 8 1 EEG recording for one animal from 0 to 8 minutes after the drug injection.

PAGE 115

115 Figure 8 2 Selected continuous absolute absorption coefficient cross section images at different time points 1 mm under the injection site of the brain during the resting state. Figure 8 3 Selected continuous absolute absorption coefficient cross section images at different time point s 1 mm under the injection site of the brain after the drug injection.

PAGE 116

116 Figure 8 4 (a) A 70s EEG signal piece that is corresponding to 1000 imaging frames of DOT. Significant changes of the spike frequencies could be observed in this piece. (b) P ower spectra of the EEG signal. (a) (b)

PAGE 117

117 Figure 8 5 Curves of the averaged HbT value ove r the same 70s as that of Fig. 8 4 Blue line represents the 2mm area of the injection site at the injection plane; red line represents the 2mm area on the plane 3mm from the injection level; green line represents the area 10mm from the injection site.

PAGE 118

118 Figure 8 6. Detailed comparison of EEG abnormalities and corresponding optical signal response. (a d) EEG signals, each represented a piece of 2s length ; (e h) corresponding response of optical signals (a) (b) (e) (f) (c) (d) (g) (h)

PAGE 119

119 CHAPTER 9 CONCLUSIONS AND FUTU RE STUDIES 9 .1 Conclusions This dis sertation has explored the abilities of diffuse optical tomography to detect multiple spatial and functional features in breast cancer and epilepsy from both the hardware and software aspects. Generally, both tissue mimicking phantom studies and in vivo st udies show that DOT is capable of identifying breast lesions and epileptic abnormalities. The versatilities of DOT are tested and proved in this study, suggesting it to be a strong image modality. W e have developed a two step PCDOT method which is able to improve the RI reconstruction qualitatively and quantitatively making PCDOT a potential ly observer independent approach for cancer diagnosis. T his method has been confirmed by phantom experiments and 42 sets of clinical data. The reflection mode based DO T system played an important role in this study. Using this system, preliminary study was conducted and we managed to prove that DOT is able to detect localized seizure onset and the measured signal shows a linear relationship with the input neural activities that similar linear system studies could be applied to analyze quantitative hemoglobin, oxygenation and blood flow information u sing multi wavelength detections. Using the multi wavelength DOT system, multiple parameters including hemoglobin and CBF were recovered through a robust reconstruction algorithm. And DOT is proved to be able to locate the seizure focus under different fu nctional parameters. Through the fast DOT imaging system, 14Hz image frames were recovered and the surround inhibition was observed, showing inner coupling of hemodynamic

PAGE 120

120 response towards neural abnormal activities The findings would serve as an important role in shaping the evolution of seizure. 9 .2 Future Studies For the breast cancer detection part, w e expect to further evaluate this method using larger scale clinical data as well as applying this method for imaging other organs/diseases. For brain ima ging, b ased on the current observations, this system also has the potential to spatiotemporally detect the earliest neurovascular changes associated with seizure anticipation and onset, including providing a new tool for real time seizure prediction based on the hemodynamic change before the seizure onset. Coupling study among the functional variables would bring insights into how these physiological features accompanied and interacted with each other during seizure activities, for instance the long standin g debate of if cerebral blood flow is able to meet the increased enormous metabolic demands during seizures; and coupling study between functional information and neural responses would help to understand mechanisms of how functional variations are associa ted with different aspects of neuronal activities including the average firing rate of all or partially of the neurons, the local field potential, the local synaptic activities and synchronized spiking activities. The current epilepsy study is limited to t he acute focus model only. While generalized seizures are more common in the actual situations, the next step of the study would be directed to the generalized seizure models. Current study have prepared hardware of data collection, reconstruction algorith m and interpretation methods. Should all these set as a robust foundation to investigate more complicated abnormal brain phenomenon. And last but not the least, DOT could be easily translated

PAGE 121

121 to human and become a useful imaging modality to dynamically ima ge seizure related activities.

PAGE 122

122 LIST OF REFERENCES 1. M. Culter Transillumination as an aid to diagonosis of breast lesions, Surg. Gynaecol Obstet. 48, 721 727(1929). 2. CH. Jones ethods of breast imaging Phys. Med. Biol., 27, 463 499 (1982) 3. B. Chance, S. Nioka, J. Zhang, E. F. Conant, E. Hwang, S. Briest, S. G. Orel, M. D. biochemical and physiological properties of breast cancers: A six year, two site d. Radiol. 12, 925 933 (2005) 4. S. Srinivasan, B. W. Pogue, S. D. Jiang, H. Dehghani, C. Kogel, S. Soho, J. J. and water concentration, oxygen saturation, and scattering mea sured in vivo by near 12354 (2003) infrared optical imaging of the breast with model adiol. 9, 186 194 (2002) 6. A. Y. Bluestone, G. Abdoulaev, C. H. Schmitz, R. L. Barbour, and A. H. Hielscher, Express 9, 272 286 (2001) in vivo : techniques and applications from animal 8. tomographic image reconstruction from ultrafast time sliced transmission ppl. Opt. 38, 4237 4246 (1999) 9. H. Jiang, K. D. Paulsen, U. L. Osterberg, B. W. Pogue, and M. S. Patterson, from near infrared frequency 2130 (1995) 10. E. Okada, M. Firbank, M. Schweiger, S. R. Arridge, M. Cope, and D. T. Delpy, infrared light propagation in a 31 (1997) 11. Y. Xu, X. Gu, T. Khan heterogeneous scattering media can be simultaneously reconstructed by use of dc 5437 (2002) ee Trans. Med. Imaging 20, 1334 1340 (2001)

PAGE 123

123 physiology: a study of biomedical intrinsic and extr Soc. London Ser. B 352, 707 716 (1997) wave 298 (1999) w 279 (1997) 16. http://www.cancer.org/acs/groups/content/@epidemiologysurveilance/documents/do cument/acspc 030975.pdf accessed 06/12 17. EA. Sickles DL. Miglioretti R. Ballard Barbash et al. Performanc e benchmarks for diagnostic mammography Radiol 235 775 790 (2005) 18. PA. Carney DL. Miglioretti BC. Yankaskas et al. Individual and combined effects of age, breast density, and hormone replacement therapy use on the accuracy of screening mammogr Ann. Intern. Med. 138, 168 175 (2003) 19. DA. Bluemke CA. Gatronis MH. Chen et al. Magnetic resoncance imaging of the breast prior to biopsy 2735 2742 (2004) atients 118 (1995) 21. A. Cerussi, D. Jakubowski, N. Shah, F. Bevilacqua, R. Lanning, A. J. Berger, D. the information cont 71 (2002) 22. B. J. Tromberg, N. Shah, R. Lanning, A. Cerussi, J. Espinoza, T. Pham, L. invasive in vivo characterization of breast tumors using photon migration spectrosco 40 (2000) 23. P. Vaupel, F. Kallinowsk oxygen and nutrient supply, 6449 6465 (1989) 24. VG. Peters DR Wyma MS Patterson, GL Frank Optical properties of normal and diseased human breast tissues in the visible and near infrared Phys. Med. Biol. 35,1317 1334 (1990) 25. AE Profio GA Navarro OW Sartorius, Scientific basis of breast diaphanography Med. Phy.16,60 65 (1989)

PAGE 124

124 53 60 (2004) i nfrared optical imaging of the breast with model 194 (2002) 28. BW Pogue, S Jiang, H Dehghani, C Kogel, S Soho, S Srinivasan, X Song, TD NIR scattering in breast tissue: analysis of intersubject variability and menstrual 552 (2004) infrared imaging using continuous, phase modulated, and pulsed light with quantitation of blood and b 29 45 (1998) 30. BJ Tromberg, O Coquoz, J Fishkin, T Phanm, ER Anderson, J Butler, M Cahn, JD invasive measurements of breast tissue optical properties using frequency domain pho B. 352, 661 668 (1997) 31. V Ntziachirstos, AG Yodh, M Schnall, B Chance, ,Concurrent MRI and diffuse Acad. Sci. USA 97, 2767 2772 (2000) 32. V. Ntziachristos AG. Yodh MD. Schnall et al. MRI guided diffuse optical spectroscopy of malignant and benign breast lesions 347 354 (2002) 33. MA. Franceschini KT. Moesta S. Fantini et al. Frequency domain techniques enhance optical mammography: Initial clinical results Prod Natl Acad Sci USA 94 6468 6473 (1997) 34. D. Grosenick KT. Moesta H. Wabnitz et al. Time domain optical of breast tumors Appl Opt. 42 3170 3186 (2003) 35. D. R. Leff, O. J. Warren, L. C. Enfield, A Gibson, T. Athanasiou, D. K. Patten, J. Res. Treat. 108, 9 22 (2008) 36. H. Jiang Y. Xu, Phase contrast imaging of tissue using near infrared diffusing light Med Phys 30 1048 1051 (2003) 37. AH. Bennett H. Osterberg H. Jupnik O. Richards, Phase microscop e: principles New York: John Wiley & Sons ( 1951 ) 38. J. Bereiter Hahn C. Fox B. Thorell, Quantitative reflection contrast microscope of living cells J Cell Biol 82, 767 779 (1979)

PAGE 125

125 39. T. Takeda A. Momose K. Hirano S. Haraoka T. Watana be Y. Itai, Human carcinoma: Early ex p erience with phase contrast X ray CT with Synchrotron radiation : Comparative specimen study with optical microscope Radiology 214 298 301 (2000) 40. J. Davis D. Gao T. Gureyev A. Stevenson S. Wilkins, Phase contrast imaging of weakly absorbing materials using hard X rays 595 598 (1995) 41. JF. G Computer Proc IEEE 71, 330 337 (1983) 42. GJ. Tearney ME. Brezinski J. Southern B. Bouma M. Hee J. Fujimoto, Determination of the refractive index of highly scattering human tissue by optical coherence tom ography Opt Lett 20, 2258 2260 (1995) 43. A. Zysk E. Chaney S. Boppart, Refractive index of carcinogen induced rat Phys. Med. Biol. 51, 2165 2177 (2006) 44. WA Hauser LT Kurland, The epidemiology of epilepsy in Rochester, Minnesota, 1935 through 1567 Epilepsia; 16:1 66 (1975) 45 Commission on classsification and Terminology of the International League Against Epilespy: Proposal for revised clinical and electroencephalographic classification of epileptic seizures. Epileps ia; 22:489 501 (1981) 46. K Goffin S Dedeurwaerdere et al. Neuronuclear assessment of patients with epilepsy ; 38:227 239 (2008) 47. EL So, Role of Neuroimaging in the management of seizure disorders Proc ; 77:1251 1264 (2002) 48. JF Annegers WA Hauser JR Lee WA Ro cca Incidence of acute symptomatic seizures in Rochester, Minnisota, 1935 1984 Epilepsia; 36:327 333 (1995) 49 JW Britton EL So Selection of antiepileptic drugs: a practical approach Mayo Clin Proc; 71:778 786 (1996) 50. F Rosenow H Luders Presurgical evaluation of epilepsy ; 124: 1683 1700 (2001) 51. CE Elger C Helmstaedter M Kurthern Chronic epilepsy and cognition Lancet Neurol; 3:663 672 (2004) 52. SF Berkovic AM Mcintosh RM Kalnin et al. Preoperative MRI predicts outcome of temporal lobectomy: an actuarial analysis Neurology 45:1358 1363 (1995) 53. RK Mosewich, EL So, et al. Factors predictive of the outcome of frontal lobe epilepsy surgery Epilepsia 41:843 849 (2000)

PAGE 126

126 54. Ch ugani HT, Shewmon DA, Shields WD, et al. Surgery for intractable infantile spasms: neuroimaging perspectives Epilepsia 34:764 771 (1993) 55. Drzezga A, Arnold S, Minoshima S, et al. 18F FDG PET studies in patients with extratemporal and temporal epilep sy: evaluation of an observer independent analysis J Nucl Med 40:737 746 (1999) 56. Breier JI, Mullani NA, Thomas AB, et al. Effects of duration of epilepsy on the uncoupling of metabolism and blood flow in complex partial seizures Neurology 48:1047 1 053 (1997) 57. Casse R, Rowe CC, Newton M, et al. Positron emission tomography and epilepsy Mol Imaging Biol 4:338 35 (2002) 58. Bonte FJ, Devous MD Sr, Stokely EM, Homan RW Single photon tomographic determination of regional cerebral blood flow in epilepsy AJNR Am J Neuroradiol 4:544 54 (1983) 59. Sadadzic S, Yuan S, Dilekoz E, Ruvinskaya S, Vinogradov SA, Ayata C, and Boas DA Simultaneous imaging of cerebral partial pressure of oxygen and blood flow during functional activation and cortical spr eading depression Applied Optics 48: D169 D177 (2009) 60. Lauritzen M Relationship of spikes, synaptic activity, and local changes of cerebral blood flow Journal of Cerebral Blood Flow and Metabolism 21: 1367 1383 (2001) 61. Annegers JF, Hauser WA, L ee JR, Rocca WA Incidence of acute symptomatic seizures in Rochester, Minnisota 1935 1984. Epilepsia 36:327 333 (1995) 62 DeSalvo MN, Schriddde U, Mishra AM, Motelow JE, Purcaro MJ, Danielson N, Bai X, Hyder F, and Blumenfeld H Focal BOLD fMRI chang es in bicuculline induced tonic clonic seizures in the rat Neuroimage 50: 902 909 (2010) 63. Salek Haddadi A, merschhemke M, Lemieux L and Fish DR Simultaneous EEG correlated ictal Fmri Neuroimage 16:32 40 (2002) 64. Schwartz TH, Bonhoeffer T (2001) In vivo optical mapping of epileptic foci and surround inhibition in ferret cerebral cortex. Nat Med 7:1063 1067 65. Bahar S, Suh M, Zhao M and Schwartz TH Intrinsic optical signal imaging of NeuroReport 17: 499 503 (2006) 66 Adelson PD, Nemoto E, Scheuer M, et al. Noninvasive continuous monitoring of cerebral oxygenation periictally using near infrared spectroscopy: a preliminary report Epilepsia 40:1484 1489 (1999)

PAGE 127

127 67. Bluestone A, Stewart M, Laske r J, Abdoulaev J, Hielscher A Three dimensional optical tomographic brain imaging in small animals, part 1: hypercapnia J Biomed 9:1046 1062 (2004) 68. Bluestone A, Stewart M, Lasker J, Lei B, Kass I, et al. Three dimensional optical tomographic brai n imaging in small animals, part 2: unilateral carotid occlusion J Biomed 9:1063 1073 (2004) 69 Culver JP, Drduran T, Furuya D, Cheung C, et al. Diffuse optical tomography of cerebral blood flow, oxygenation, and metabolism in rat during focal ischemia J Cereb Blood Flow Metab 23:911 924 (2003) 70 Culver JP, Siegel AM, Boas DA Three dimensional diffuse optical tomography of fore paw stimulation in a rodent Opt Lett 28:2061 2063 (2003) 71. Gibson AP, Austin T, Everdell NL, Schweiger M, et al. Three dimensional whole head optical tomography of passive motor evoked responses in the neonate J Neuroimaging 30:521 528 (2006) 72. Hintz SR, BEnaron DA, Siegel AM, Zourabian A, et al. Bedside functional imaging of the premature infant brain during pa ssive motor activation J Perinat Med 29:335 343 (2001) 73 Wang Q, Liang X, Liu Z, Carney P, Jiang H Visualizing localized dynamic changes during epileptic seizure onset in vivo with diffuse optical tomography Medical Physics 35:216 224 (2008) 74. Og awa B, Tank DW, Menon R, Ellermann JM, et al. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging Proc Natl Acad Sci USA 89:5951 5955 (1992) 75. Kwong KK, Belliveau JW, Chesler DA, et al. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation Proc Natl Acad Sci USA 89:5675 5679 (1992) 76. Joseph D, Huppert TJ, Franceschini MA, Boas DA Diffuse optical tomography system to image brain activation wit h improved spatial resolution and validation with functional magnetic resonance imaging App Opt 45:8142 8151 (2006) 77 Aghakhani Y, Bagshaw AP, Benar CG, Hawco C, Andermann F, Dubeau F and Gotman J fMRI activation during spike and wave discharges in i diopathic generalized epilepsy. Brain 127(5): 1127 1144 (2004) 78. A. Bluestone, M. Stewar J. Lasker J. Abdoulaev A. Hielscher Three dimensional optical tomographic brain imaging in small animals, part 1: hypercapnia J Biomed 9, 1046 1062 (2004)

PAGE 128

128 79. DE. Kuhl J. Engel ME. Phelps AP. Kowell Epileptic patterns of local cerebral metabolism and perfusion in man: investigation by emission computed tomography of 18F flu orodeoxyglucose and 13N ammonia Trans Am Neurol Assoc 103 52 53 (1978) 80. M. Ingver B. Nillsson and B. K. Siesjo Local cerebral blood flow in the brain during bicuculline induce seizures and the modulating influence of inhibition of prostaglandin sysnthesis Acta Physiol Scand 111 205 212 (1981) 81. X. Liang, Q. Zhang, H. Jiang, Quantitative reconstruction of refractive index distribution and imaging of glucose concentrantion by using diffusing light, App. Opt. 45, 8360 8365(2006) 82. Duncan JS Imaging and epilepsy 339 377 (1997) 83. Arridge SR, Schweiger M,Hiraoka M, Delpy DT, A finite element approach for modeling photon transport equation approximation, Med. Phys. 20, 299 309 (1993) 84. Jiang H, Xu Y, Iftimia N, Experimental three dimensional optical image reconstruction of heterogeneous turbid media from continuous wave data, Opt. Express 7, 204 209 (2000) 85. Phase contrast a diol. 15,859 866(2008) 86. H. Jiang, T he diffusion approximation for t urbid media with a spatially varying refractive index Proc Adv Opt Imaging Photon Migration Opt Soc Am 366 368 (2000) 87. T. for scatte Opt. 5, 137 141 (2003) 88. B. W. Pogue Opt. 38, 2950 2961 (1996) 89. Z. Yuan, Q. Zhang, E. S. Sobel, H. Jiang, Tomographic x ray guided three dimensional diffuse optical Opt. 13(4) 047804 (2008) 90. Today 48, 34 40 (1995) 91. B. J. Tromberg, N. Shah, R. Lanning, A. Cerussi, J. Espinoza, T. Pham, L. invasive in vivo characterization of breast tumors using photon migration spectro 40 (2000)

PAGE 129

129 92. B. W. Pogue, S. P. Poplack, T. O. McBride, W. A. Wells, O. K. S. Osterman, U. L. near infrared spectroscopy: pilot results in the brea 266 (2001) 93. A. Cerussi, D. Jakubowski, N. Shah, F. Bevilacqua, R. Lanning, A. J. Berger, D. Opt. 7(1), 60 71 (2002) 94. J. R. Mourant, A. H. Hielscher, A. A. Eick, T. M. Johnson, and J. P. 374 (1998) 95. Saunders, Philadelphia, Pennsylvania (1994) diffuse optical tomography with absorp App. Opt. 46 (34), 8229 8236 (2007) migration methods for particle sizing in concentrated 1744 (1998) 98. C. Li, H. Zhao, B. Anderson, H. Jiang, Mul tispectral breast imaging using a ten wavelength, 64x64 source/detector channels silicon photodiode based diffuse optical tomography system, Med. Phys. 33627 636 (2006) 99. detection diffuse optical Rev. Sci. Instr. 74,2836 2842(2003) 100. infrared optical imaging of the breast with model 194(2002) 101. JJ Casciari, S and metabolism with oxygen concentration, glucose concentration, and 394(1992) 102. contrast imaging of tissue using near infrared diffusing 1051(2003) 103. Pei Y, Graber HL and Barbour RL Influence of systematic errors in reference states on image quality and on stability of derived information for dc optical imaging Applied Optics 40(31): 575 5 5769 (2001) Imaging, 27(8):1163 74 (2009)

PAGE 130

130 Statistical parametric maps in functional imaging: a Human Brain Mapping, 2:189 210 (1995) 106. Carmichael DW, Hamandi K, Laufs H, Duncan J, Thomas DL and Lemieux L (2008) An investigation of the relationship between BOLD and perfusion signal changes during epileptic generalized sp ike wave activity. Magnetic Resonance Imaging 26: 870 873 107. Worsley KJ, Liao CH, Aston J, Petre V, Duncans GH, Morales F and Evans AC (2002) A general statistical analysis for fMRI data. NeuroImage 15: 1 15 108. Liu Z, Rios C, Zhang N, Yang L, Chen W an d He B (2010) Linear and nonlinear relationships between visual stimuli, EEG and BOLD fMRI signals. NeuroImage 50:1054 1066 109. DeSalvo MN, Schriddde U, Mishra AM, Motelow JE, Purcaro MJ, Danielson N, Bai X, Hyder F, and Blumenfeld H Focal BOLD fMRI changes in bicuculline induced tonic clonic seizures in the rat Neuroimage 50: 902 909(2010) 110. Steriade M and Contreras D Spike wave complexes and fast components of cortically generated seizure. I. Role of neocortex and thalamus J ournal of Neurophysiology 80(3):1439 1455 (1998) 111. Theodore WH, Balish M, Leiderman D, Bromfiled E, Sato S, and Herscovitch P Efflect of seizures on cerebral blood flow measured with 15O h2O and positron emission tomography Eplilepsia 37: 796 802 (1 996) tomography system for in vivo three App. Opt. 51(16):3461 3469 (2012) 113. ht tp://omlc.ogi.edu/spectra/hemoglobin/ accessed 06/2012 114. Mintun, M. A., Raichle, M. E.,. Mart Brain oxygen utilization measured with O 15 radiotracers an J. Nucl. Med. 25, 177 187 (1984). 115. What we can do a Nature 453 869 878(2008). 116. Fabricius, M., Akgoren, N Laminar analysis of cerebral blood flow in cortex of rats by laser Do ppler flowmetry, a pilot J. Cereb. Blood Flow Metab 17 1326 1336 (1997) 117. Cheung, C., Culver, J.P., Takahashi, K. In vivo cerebrovascular measurement combining diffuse near infrared absorption and Phys. Med. B iol 46 2053 2065(2001)

PAGE 131

131 2010 Technical Digest of OSA Biomedical Optics and 3D Imaging Topical Meeting ( Digital Holography and Three Dimensional Imaing), Miami, Florida, USA, April 12 14, 2010. BWA3 119. Herman P, Sanganahalli BG, Bl umenfeld H, Hyder F, 'Cerebral oxygen demand for short li ved and steady state events.'J. Neurochem.;109 Suppl 1:73 9 (2009) 120. Logothetis NK, Pauls J, A ugath M, Trinath T, Oeltermann A 'Neurophysiological investig ation of the basis of the fMRI signal', Nature ;412(6843):150 7 (2001) 121. Martindale J, Mayhew J, Berwick J, Jones M, Martin C, Johnston D, Redgrave P, Zhe ng Y, 'The he modynamic impulse response to a single neural eve nt', J. Cereb. Blood Flow Metab. ;23(5):546 55 (2003) 122. Arthurs OJ, Williams EJ, Carpenter TA, Pickard JD, Boniface SJ Linear coupling between functional magnetic resonance imaging and evoked potential amp litude in human somatosensory cortex.' Neuroscience;101(4):803 6 (2000) 123. Ou W, Nissil I, Radhakrish nan H, Boas DA, Hmlinen MS, Franceschini MA.' Study of neurov ascular coupling in humans via simultaneous magnetoencephalograp hy and diffuse optical imaging acquisition', Neuroimage. ;46(3):624 32. Epub (2009) 124. Jacco A de Zwart, Peter v an Gelderen, J Martijn Jansma, Masaki Fukunaga, Marta Bianciardi, and Jeff H Duyn, 'Hemodynamic Nonlinearities Affect BOLD fMRI R esponse Timing and Amplitude', Neuroi mage.; 47(4): 1649 1658 (2009) 125. Anna Devor*, 1, Andrew K. Dunn1, Mark L. Andermann1, 2, Istvan Ulbert1, 3, David A. Boas1 a nd Anders M. Dale, 'Coupling of Total Hemoglobin Concentr ation, Oxygenation, and Neural Activity in Rat Somatosensory Co rtex,' Neuron, Volume 39, Issue 2, 353 359 (2003) 126. Hernandez, M.J., Brennan, R.W blood flow 154(1978) 127. Prince DA, Wilder BJ, 'Control mechanisms in cortical epileptogenic foci. "surro und" in hibition.' Arch Neurol 16:194 202 (1967) 128. Devor A, Tian P, Nishimura N, Teng IC, Hillman EM, Narayanan SN, ULbert I, Boas DA, Kleinfeld D, Dale AM, Suppressed neuronal activity and concurrent arteriolar vasoconstriction may explain negative b lood oxygentaion level dependent signal.' J. Neurosci. 27:4452 4459 (2007) 129. Schwartz TH, Bonhoeffer T, 'In vivo optical imaging of epileptic foci and surround inhibition in ferret cerebral cortex', Nat. Med. 7:1063 1067 (2001)

PAGE 132

132 130. Hirase H, Creso J, B uzsaki G, Capillary level imaging of local cerebral blood flow in bicuculline induced epileptic foci.' Neuroscience 128:209 216 (2004a) 131. Zhao M, Ma H, Suh M, Schwartz TH, 'Spatiotemporal dynamics of perfusion and oximetry during ictal discharges in t he rat neocortex.' J. Neurosci. 29:2814 2823 (2009) 132. Zhao M, Nguyen J, Ma H, Nishimura N, Schaffer CB, Schwartz TH, 'Preictal and ictal neurovascular and metabolic coupling surrounding a seizure focus,' J. Neurosci. 31(37):13292 13300 (2011)

PAGE 133

133 BIOGRAPHICAL SKETCH Ruixin Jiang earned her B.S. in e lectrical e ngineering from University of Electronics and Technology of China in 20 06. She earned her M.S. in biomedical e ngineering from University of Florida in 2009. She entered the Ph. D. program in b iomedical e ngineering University of Florida in 2006 and she received her Ph.D. in the fall of 2012.