RESPIRATORY MOTION MANAGEMENT FOR RADIATION THERAPY By ASHLEY GALE SMITH 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 2015
Â© 2015 Ashley Gale Smith
To my loving husband Jason and loyal dog Callie
4 ACKNOWLEDGMENTS I would like thank each of the faculty members serving on my superviso ry committee, including committee chair, David Hintenlang , and committee members, Kathleen Hintenlang, Hans van Oostrom , Christopher Serago, and Nicholas Bacon . Their input and advice provided me with valuable guidance which helped keep my research focused and moving forward. I would like to thank Siyong Kim from Virginia Commonwealth University. His instruction and ideas at the beginning of my studies provided a great foundation for my research, and I have appreciated his continued help and support. I wo uld like to thank Robert Pooley from Mayo Clinic for the use of his equipment and for working with me after hours to complete MR testing. His help and advice have been greatly appreciated. I would also like to thank Mark Jenkins from Mayo Clinic whose en gineering skills were of great help many times throughout my time as a graduate student, and for making the process enjoyable. I would also like to give a special thank you to Christopher Serago from Mayo Clinic for encouraging me to pursue a doctoral degr ee. His advice and guidance during my studies at the University of Florida as well as throughout my entire career as a medical physicist have been invaluable. This kind of support and encouragement from a boss is rare to find. He is always going above a nd beyond to help further my career and always has my best interest at heart. His patience and understanding made completing a doctoral degree while working full time happen as quickly and smoothly as possible. Finally, I would like to thank my family and friends. I would like to give thanks to my mom and dad for always encouraging me to go after my dreams and instilling in me the work ethic and values I needed to succeed in life. I would like to give thanks to my
5 sisters , Shannon and Heather, for always b elieving in me and encouraging me in my academic endeavors . I would like to thank my sweet Callie for her endearing loyalty and companionship. I could always count on her to listen to my presentations, calm my nerves, and provide comic relief. And last but not least I want to thank my husband Jason for his constant love, encouragement, and support. He has always supported me in everything I do. His effort to make my life easier during my graduate education was amazing. I love his positive attitude an d sense of humor which inspired me to keep going . I appreciate everything he does for me whether small or large to show me how much he loves me.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 Current State of Motion Management in Rad iation Therapy ................................ ... 16 ABC ................................ ................................ ................................ .................. 16 VBH ................................ ................................ ................................ .................. 17 DIBH ................................ ................................ ................................ ................. 17 Respiratory Gating ................................ ................................ ........................... 18 Motion Encompassing Methods ................................ ................................ ....... 19 FSB with Abdominal Compression ................................ ................................ ... 20 Real Time Tumor Tracking ................................ ................................ ............... 20 Improving Motion Management in Radiation Therapy ................................ ............ 21 Activ e Approach ................................ ................................ ............................... 21 Passive Approach ................................ ................................ ............................ 23 Specific Aims of this Research ................................ ................................ ............... 25 2 TEMPERATURE BASED OPTICAL FIBER RESPIRATION SENSOR .................. 27 Bac kground ................................ ................................ ................................ ............. 27 Materials and Methods ................................ ................................ ............................ 27 Materials and Experimental Setup ................................ ................................ .... 29 Respiratory Cycle Tracking with Volunteer Subjects ................................ ........ 29 CT, Linear Accelerator, and MR Compatibility ................................ .................. 30 Results ................................ ................................ ................................ .................... 32 Test of Respiratory Cycle Tracking with Volunteer Subjects ............................ 32 Test of CT, Linear Accelerator, and MR Compatibility ................................ ...... 32 Discussion ................................ ................................ ................................ .............. 33 3 DOSIMETRIC EVALUATION OF THE INTERPLAY EFFECT WITH FFF BEAMS ................................ ................................ ................................ ................... 43 Background ................................ ................................ ................................ ............. 43
7 Materials and Methods ................................ ................................ ............................ 45 Results ................................ ................................ ................................ .................... 46 Discussion ................................ ................................ ................................ .............. 47 4 USE OF FFF BEAMS FOR OFF AXIS TARGETS ................................ ................. 52 Background ................................ ................................ ................................ ............. 52 Materials and Methods ................................ ................................ ............................ 54 Lateral Target, Off Axis Geometry Evaluation ................................ .................. 54 Mechanical Rotation Induced Dosimetric Uncertainty Evaluation ..................... 54 Statistical Analysis ................................ ................................ ............................ 55 Results ................................ ................................ ................................ .................... 56 Lateral Target, Off Axis Geometry Evaluation ................................ .................. 56 Mechanical Rotation Induced Dosimetric Uncertainty Evaluation ..................... 57 Discussion ................................ ................................ ................................ .............. 58 5 CONCLUSIONS ................................ ................................ ................................ ..... 71 Results of This Work ................................ ................................ ............................... 71 Opportu nities for Future Work and Development ................................ .................... 73 LIST OF REFERENCES ................................ ................................ ............................... 74 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 80
8 LIST OF TABLES Table page 2 1 Fiber optic temperature sensor signal during CT scanning. ............................... 36 2 2 Fiber optic temperature sensor signal in high energy linear accelerator environment. ................................ ................................ ................................ ....... 36 2 3 Respiration signal recorded by fiber optic temperature sensor in MR environment. ................................ ................................ ................................ ....... 37 2 4 MR American College of Radiology Quality Assurance. ................................ ..... 37 2 5 CT American College of Radiology Quality Assurance. ................................ ...... 37 3 1 Delivered vs planned dose distribution for 1000 trials of randomly selected starting phases. ................................ ................................ ................................ .. 49 3 2 Dose difference for delivered vs planned dose for central and peripheral target points. ................................ ................................ ................................ ....... 49 4 1 L ateral target, off axis geometry evaluation results for 6 MV. ............................. 62 4 2 Lateral target, off axis geometry evaluation results for 10 MV. ........................... 62 4 3 Results of prescription isodose surface coverage, conformality indices, and lung V20 for OAG and CAG. ................................ ................................ .............. 63 4 4 Absolute difference of mechanical rotation and original plan results for O AG and CAG. ................................ ................................ ................................ ............ 64
9 LIST OF FIGURES Figure page 2 1 Fiber optic temperature sensor in plastic holder. Photo courtesy of author. ...... 38 2 2 Mask system consists of plastic holder attached to disposable mask. Photo courtesy of author. ................................ ................................ .............................. 38 2 3 Temperature based respiration system and RPM e xternal marker used to simultaneously record volunteer breathing cycle with FSB. Photo courtesy of author. ................................ ................................ ................................ ................ 39 2 4 Sample breathing data (both sensors tracked cycle). ................................ ......... 40 2 5 Sample breathing data (RPM data not usable for 4D CT). ................................ . 40 2 6 Volunteer breathing cycles in MR environment. ................................ ................. 41 2 7 MR low contrast object detectability with no sensor (left) and with sensor (right). ................................ ................................ ................................ ................. 42 3 1 2D ionization chamber array on motor driven motion platform (Respiratory Gating Platform; Standard Imaging, Middleton, WI). Photo courtesy of author. ................................ ................................ ................................ ................ 50 3 2 Points selected at the periphery of the target region (OmniPro ImRT; IBA, Bartlett, TN). ................................ ................................ ................................ ....... 51 4 1 CAG and OAG for a laterally located target. ................................ ....................... 65 4 2 Mechanical rotation induced dosimetric uncertainty evaluation for 5 cases. Representative axial slice for OAG (left) and CAG (right). ................................ .. 66 4 3 Prescription Isodose Surface Coverage for OAG and CAG for the original plans and mechanical rotation plans. Points and lines for the same pati ent are shown in the same color. ................................ ................................ .............. 67 4 4 CI 100 for OAG and CAG for the original plans and mechanical rotation plans. Points and lines for the same patient are shown in the same color. ........ 68 4 5 CI 50 for OAG and CAG for the original plans and mechanical rotation plans. Points and lines for the same patient are shown in the same color. ................... 69 4 6 Lung V20 for OAG and CAG for the original plans and mechanical rotation plans. Points and lines for the same patient are shown in the same color. ........ 70
10 LIST OF ABBREVIATIONS 2D Two dimensional 3D Three dim ensional 4D Four dimensional ABC Active breathing control ACR American College of Radiology BOT Beam on time CAG Central axis geometry CT Computed tomography CI Conformality index CI 50 Ratio of 50% prescription isodose volume to target volume CI 100 Ratio of prescription isodose volume to target volume DCAT Dynamic conformal arc therapy DIBH Deep inspiration breath hold FF Flattening filter FFF Flattening filter free FSB Forced shallow breathing HI Homogeneity index IGRT Image guided radiat ion therapy IMRT Intensity modulated radiation therapy MR Magnetic resonance MLC Multi leaf collimator MU Monitor unit s MV Megavoltage OAG Off axis geometry
11 PIU Percentage integral uniformity PTV Planning target volume QA Quality assurance RPM Re al time Position Management RTOG Radiation Therapy Oncology Group SBRT Stereotactic body radiation therapy SI Superior inferior SNR Signal to noise ratio SRS Stereotactic radiosurgery SSD Source to surface distance TG 76 Task Group No. 76 VBH Volun tary breath hold VMAT Volumetric modulated arc therapy
12 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 RESPIRAT ORY MOTION MANAGEMENT FOR RADIATION THERAPY By Ashley Gale Smith August 2015 Chair: David Hintenlang Major: Biomedical Engineering Radiotherapy motion management is important when tumor movement is present due to respiration. One approach to motion m anagement is respiratory gating, which synchronizes radiation delivery with respiration. A temperature based respiration sensor was developed that does not require abdominal movement like commonly used external fiducial markers. The sensor tracks breathi ng cycles by measuring the temperature of inspiratory and expiratory air, so it can be used with forced shallow breathing (FSB) or thermoplastic body mask immobilization. The temperature sensor was used to track breathing cycles for five volunteers using FSB, and compared to simultaneously recorded signals from an external marker. Temperature sensor performance was tested in CT, MR, and linear accelerator environments. The sensor effect on image quality was evaluated for CT and MR. The temperature sensor successfully recorded breathing cycles for all volunteers, while the external marker had one failure. Temperature sensor signal in CT and MR environments was similar to background signal; however it is recommended the sensor be placed outside the linear accelerator radiation field. Temperature signals were
13 obtained without deterioration of CT or MR images. These results indicate this device is suitable for respiratory gating, and an improvement to external markers. Another approach to motion management is to use flattening filter free (FFF) beams to reduce beam on time, which aids in breath hold techniques and reduces changes over time in breathing patterns. The interplay effect with FFF beams was investigated using an ionization chamber array on a movi ng platform to simulate respiratory motion. Measured dose distributions were compared to planned dose for patient plans using gamma analysis and percent of pixels within 5% and 10% of planned dose. The impact of the interplay effect was observed on dose distributions; however deviations were not influenced by high FFF dose rate. FFF plans were also evaluated for lateral targets with homogeneity and conformality indices to compare central and off axis geometry (OAG). It was shown FFF beams can be used in OAG without dosimetric compromise. This research contributed to respiratory management techniques with an improvement to respiratory gating sensors and by verifying FFF has a role in motion management.
14 CHAPTER 1 INTRODUCTION Radiation therapy is the trea tment of disease, especially cancer, using high intensity ionizing radiation. R adiation kills cancer cells by damaging their genetic material. Radiation can damage normal cells in addition to cancer cells, so the radiation treatment must be carefully pla nned. This research focuses on external beam radiotherapy, which is the most common form of radiotherapy. External beam radiotherapy typically uses high energy photons in the megavoltage range produced by a linear accelerator to treat cancer. The goal in radiation therapy is to give a high radiation dose to the target while l imiting dose to normal tissue. Localization of the tumor is important to achieve the treatment intent and limit toxicity. The first step in treatment planning is a simulation in whi ch a computed tomography (CT) scan is taken of the area to be treated. This process is described in detail by Coia et al . . During the simulation the patient is scanned in the same position and immobilization that they will be in for the treatment. T he patient is immobilized to reduce motion and to aid in reproducibility of the setup. The center of the treatment area, or isocenter, is defined and marked as a reference point. CT images are then used to determine the geometry of the radiation beams an d to calculate the plan dose. The technology and integration of advances in radiation therapy continues to evolve into clinical practice. Examples of these advances are dynamic conformal arc therapy (DCAT), volumetric modulated arc therapy (VMAT), and i mage guided radiotherapy (IGRT). The DCAT technique shapes the radiation field with a dynamic multi leaf collimator (MLC) around the tumor while rotating the gantry around the
15 patient. VMAT is similar to intensity modulated radiation therapy (IMRT), whic h uses a large number of small beamlets created by the MLC to modulate the dose to the target and avoid critical structures. Both VMAT and IMRT provide more conformal dose distributions compared to conventional therapies. During VMAT delivery the gantry r otates continuously around the patient, unl ike IMRT which uses multiple beams with static gantry angles . Also, during VMAT delivery the dose rate, MLC leaf moving speed, and gantry rotation speed vary . IGRT is the use of two dimensional (2D) or three dim ensional (3D) imaging just prior to treatment to pinpoint the treatment area. The common goal of all these techniques is better conformation of the dose distribution to the intended target volume and decreased toxicity of the treatment to healthy tissue . New technologies have permitted dose escalation and improved tumor control allowing for the use of new treatment techniques such as stereotactic body radiation therapy (SBRT). SBRT is a type of radiation therapy in which a few fractions of very high dose s of radiation are delivered to the tumor. The benefits for using SBRT in lung cancer treatment and technical details are discussed by Timmerman et al . [ 2, 3 ]. One limit to optimal radiotherapy treatments is the positional uncertainty of the target volu is a contributing element to this uncertainty. This positional uncertainty has required adequate margins to encompass the target range of motion which results in a large volu me of irradiated tissue and potentially increased toxicity. Respiratory motion management is critical in regions of the body where respiration causes tumor movement, including most tumor sites in the thorax and abdomen. Successful respiratory motion mana gement can result in decreasing the positional uncertainty
16 caused by respiration, reducing the margin required to encompass the target, and minimizing the volume of normal tissue irradiated. The amount a tumor moves due to respiration can vary widely, whic h has been demonstrated in a number of motion studies [ 4 10 ] and summarized by the American Association of Physicists in Medicine Radiation Task Group No. 76 (TG 76) [ 11 ]. Lung tumor motion data has shown a range of 0 to 34 mm [ 11 ]. Barnes et al . found t he average motion of lower lobe tumors is greater than middle, upper, or mediastinal tumors [ 9 ]. Also, abdominal organ motion is primarily in the superior inferior (SI) direction [ 5, 6 ]. However, it is also well known that individual motion may not follo w these generalizations. Stevens et al . found that tumor motion in the SI direction varied from 0 to 22 mm with no correlation to tumor size, location, or pulmonary function [ 10 ]. These studies suggest that individual assessment of respiratory motion is beneficial. Current State of Motion Management in Radiation Therapy There are a number of strategies that have been suggested to account for , or control respiratory motion including : active breathing control ( ABC ), voluntary breath hold ( VBH ), deep ins piration breath hold ( DIBH ), respiratory gating, motion encompassing methods, forced shallow breathing (FSB) with abdominal compression, and real time tumor tracking [ 11 ]. A brief description of each of these techniques follows. ABC ABC is a technique tha airflow. It has been described in detail by Wong et al . [ 1 2 ]. The airflow may be the CT simulation scan and again for subsequent treatments. During treatment, the
17 Ideally, t he duration of the respiratory cessation is chosen to be comfortable for the patient, typically 15 t o 20 seconds. All breath hold techniques minimize CT motion artifacts and the motion of the tumor during treatment. Treatment margins can be reduced b ecause respiration motion is lessened. A drawback of t he ABC technique however , is discomfort for the p atient due to the ventilator and forced airflow interruption. Also, this technique may not be suitable for patients with impaired respiratory function. VBH Like ABC, the VBH technique lessens the effects of respiration by only delivering radiation while t he patie nt is holding his or her breath. In contrast to ABC, VBH allows the patient to control the breath hold without the use of a ventilator. This technique is described by Kim et al . [ 1 3 ]. For all breath hold techniques the patient commonly receives training prior to simulation and treatment. The patient holds his or her breath at a predetermined phase of the breathing cycle, and the treating radiation therapist enables the accelerator beam on in conjunction with the patient breathing. If the patien t is unable to continue holding their breath they may terminate the radiation at any time. It is assumed that the breath hold and relevant internal anatomical organs and structures are reproducible from one treatment to the next. DIBH The DIBH technique requires that the patient reproduce the same deep inspiration breath hold during the CT simulation scan and while the treatment beam is on. A description of this technique is given by Hanley et al . [ 1 4 ]. Lung inflation levels are monitored by a spiromete r. The patient breathes through a mouthpiece connected
18 to the spirometer, and a nose clip is often used to prevent nose breathing with this technique. DIBH is more accurate than VBH because it uses spirometer monitoring. However, it may cause more disco mfort for the patient and b ecause patient compliance is required this technique may not be appropriate for all patients. Respiratory Gating Unlike breath hold techniques which require patient cooperation, respiratory gated treatments may be performed while the patient is breathing normally. The clinician has the option of using deep breathing or breath hold patterns during respiratory gating. This technique is described in detail by Pan et al .  and Mageras and Yorke [ 16 ]. For this technique the patie nt breathing cycle is tracked using a respiration monitor which receives either an external respiration signal or signal from internal fiducial markers. For simulation, a four dimensional ( 4D ) CT scan is acquired which takes into account movement of the tumor volume over time. Gating parameters are determined prior to the scan based on the respiratory signal. These parameters include displacement/phase, inhale/exhale, and duty cycle. The respiratory gating system then sends a trigger to the CT scanner to acquire a CT slice once per breathing cycle . For treatment, th e radiation beam is triggered during a phase of respiration determined during the simulation. Advantages of respiratory gating include the option to allow the patient to breathe normally, r educed CT motion artifacts, reduced respiratory motion during treatment, and reduced m argins. A disadvantage is that it prolongs the treatment time . Also, this technique require s gating hardware and software and additional training for radiation therapy staff.
19 Motion Encompassing Methods Motion encompassing methods are necessary when other methods that explicitly account for respiratory motion are not used during treatment. This method estimates the mean position and range of motion of the tumor during C T simulation so that this information can be used during planning. Motion encompassing methods are useful with IGRT methods that do not account for respiration so that the pre treatment imaging will most closely mat ch the images from simulation. A disadv antage of motion encompassing methods is that the radiation dose to the patient from the required imaging procedures will be greater than standard CT simulation by a factor of 2 to 15 if no efforts are made to reduce the dose [ 11 ]. There are three techniqu es that can be employed during CT simulation to estimate the range of tumor motion. First, slow CT scanning can be used to record multiple respiration phases per slice [ 17 19 ] . From this set of images a tumor motion encompassing volume can be determined. The disadvantage of this technique is the loss of spatial resolution in the CT set due to motion blurring. Second, the inhale and exhale breath hold CT method can be used, which his or her breath reproducibly [ 20, 21 ] . Two scans are taken, one while the patient holds their breath at inhale, the other at exhale. The CT scans are fused to give a tumor motion encompassing volume. This method reduces the blurring caused during slow scanning, but it will prolong the C T scanning time. Also, the tumor motion encompassing volume obtained using breath hold s during simul ation may not correlate to that of the free breathing condition during treatment. Third, respiratory gating with 4D CT can be used to determine the range of tumor motion and produce a tumor motion -
20 encompassing volume [ 15, 22 25 ] . This technique is the most accurate, but also requires the necessary equipment and training that respiratory gating entails. FSB with Abdominal Compression FSB is a technique u sed to decrease tumor motion and cause it to be more reproducible. FSB uses a compression device that is pressed against the abdomen to reduce diaphragm movement during respiration. The accuracy and reproducibility of this technique has been studied [ 26 28 ]. Negoro et al . found that use of a pressure plate reduced tumor moti on for 10 out of 11 patients [ 26 ]. For these patients, the range of tumor motion was 8 to 20 mm (12.3 mm mean) with no abdominal compression and 2 to 11 mm (7.0 mm mean) with compres sion. Real Time Tumor Tracking Real time tumor tracking methods dynamically reposition the radiation beam to follow the position of the tumor. A review of real time tumor tracking principles is found in TG 76 [ 11 ]. A number of different methodologies h ave been suggested for real time tumor tracking. This technique can use MLC beam tracking, a linear accelerator with a robotic arm, or treatment couch based motion compensation. An example of a real time tumor tracking system is the Synchrony Respiratory Tracking System integrated with the CyberKnife robotic linear accelerator (Accuray Incorporated, Sunnyvale, CA). The advantage of real time tumor tracking is that it eliminates a tumor motion margin while maintaining a 100% duty cycle for dose delivery. For successful real time tracking there must be a mechanism for detecting the tumor position. There are four methods that have been used to accomplish this: real time imaging of the tumor, usually with fluoroscopy; real time imaging of implanted fiducial markers in the tumor; tumor position prediction based on surrogate breathing
21 signals; and non radiographic tumor tracking, such as an implanted radiofrequency device. A difficulty with real time tumor tracking that all of these methods have in common is compensating for time delays in the beam positioning response. Attempts have been made to predict the tumor position in advance so that the beam can be synchronized with the tumor position in real time, however it is a complicated challenge due to fluctua tions in a typical human breathing cycle. Improving Motion Management in Radiation Therapy Active Approach A ctive approach es to motion management are direct tumor motion monitoring and active beam targeting, for example real time tumor tracking or the use of respiratory gating to synchronize imaging and delivery with respiration. Currently the most commonly used commercially available respiratory gating system is the Varian Real time Position Management (RPM) system (Varian Medical Systems, Palo Alto, CA). This system uses a reflective box that serves as an external fiducial marker that is chest or gating signal from the external marker can be compromised and at times is unusable w hen respiratory abdominal movement is limited. As part of this work, the number of times the RPM system was unable to track a patient breathing cycle was recorded for respiratory gating patients at Mayo Clinic Florida . It was found that over a 14 month p eriod the RPM system failed in 11.5% of respiratory gating patients due to limited motion . The Mayo Clinic Florida experience shows that a minimum gating motion range of 3 mm is needed for the RPM system to track a breathing cycle. There are several rea sons the gating motion range may not be large enough to be usable by the RPM system. Some patients are not respiratory gating candidates
22 because there is not enough abdominal movement during normal breathing to be accurately tracked by the RPM system. Th is is especially true in patients whose pulmonary function is compromised due to disease. Also, respiratory gating is often used along with FSB with abdominal compression. While compression can be beneficial to the patient, when it is used with respirato ry gating, the external gating signal can be compromised. With the FSB technique, there is often very little movement of the external marker, which limits the ability of the RPM system to track the breathing cycle. Additionally, thermoplastic body masks are sometimes used t o immobilize patients. This type of immobilization cover s d prevents movement of an external marker . External markers cannot be used to track the breathing cycle when using thermoplastic body masks, so an alter nate tracking method is needed . Respiratory gating could be improved by the development of a breathing cycle sensor for radiotherapy that is not subject to these limitations. Respiration sensor s for radiotherapy should also be compatible in a magnetic r esonance ( MR ) environment to be consistent with the future direction of radiotherapy incorporating MR into simulators and treatment delivery systems . The RPM system is not compatible in an MR environment due to the necessary hardware. Currently, CT is th e standard imaging used for radiotherapy simulation and treatment planning. However, MR imaging is desirable in many cases of radiotherapy because it allows superior contouring of the tumor and normal structures compared with CT due to its excellent soft tissue contrast and multi planar imaging capabilities. MR is currently used in radiotherapy treatment planning along with CT, usually via fusion of the MR im age set to the CT image set. MR
23 has not been used as the sole imaging modality for planning and d ose calculation because it lacks electron density information for dose calculation and also has documented image distortion issues. Therefore, it is common process to register or fuse the MR data set to the CT data set for treatment planning. However, th e fusion process increases the uncertainty of the treatment plan, and requires additional time and effort. Currently there are a number of developments in MR which could increase the scope of MR for use in radiotherapy treatment planning. Due to the grow ing interest in MR simulation and inclusion in treatment delivery systems , there is a need for developing a radiotherapy respiration sensor that is compatible with MR. Passive Approach P assive approach es to respiratory motion management include ABC, VBH, D IBH, motion encompassing methods, and FSB. Passive methods can be improved by reducing the time the radiation beam is on or beam on time ( BOT ). Reducing BOT aid s in the use of breath hold techniques, by enabling entire beam delivery within a short number of breath holds [ 29 ] . Also, reducing BOT can reduce changes over time in motion amplitude, respirator y period, and baseline location from one breathing cycle to another. A baseline shift is a period of irregular motion involving substantial displacement in the mean position of the tumor. This shift can be temporary or permanent. The likelihood of a bas eline shift increases with BOT. Zhao et al . found 69% of treatments exhibited increased baseline shifts with length of treatment time [ 30 ] . Also, the likelihood of intrafraction patient movement increases with BOT ]. This is especially important in IM RT and VMAT treatments which have longer beam delivery times than conventional radiotherapy. Using decreased BOT to reduce uncertainty in
24 patient movement and potenti al for baseline shifts allows for tighter margins to reduce the volume of normal tissue i rradiated . One method to reduce BOT is with the use of FFF beams which employ a high dose rate by removing the flattening filter ( FF ) from the path of the beam [ 29 , 32 36 ] . Radiation is generated in a linear accelerator by using high frequency electroma gnetic waves to accelerate electrons through a linear waveguide. Details of linear accelerator structure and design have been described by Karzmark et al . [ 37 ]. For photon mode, the electron beam then strikes a target to produce x rays. The x ray intens ity is peaked in the forward direction b ecause linear accelerators produce ele ctrons in the megavoltage range . A flattening filter of varying thickness across the beam is placed in the path of the x rays . The flattening filter, usually made of lead, prod uce s an almost uniform x ray fluence over the collimated radiation field. By removing the FF from the path of the beam the dose rate increases significantly along the central axis of the beam , and FFF beams can deliver radiation dose in a fraction of the time required by FF beams. As the clinical use of stereotactic radiosurgery ( SRS ) and SBRT has increased, there has been increasing i nterest in the use of FFF beams . Dose homogeneity is not as important a consideration for SRS and SBRT as it is with larg e fields or structures far off axis. This is because the degree of non uniformity of FFF fields is more prominent in larger field sizes and can be negligible in the small fields typically used in SRS and SBRT. Also, BOT reduction is especially beneficial to SRS and SBRT because both have longer delivery times than conventional therapy due to the large number of monitor units (MU) required to deliver high dose per treatment fraction.
25 Specific Aims of this Research The purpose of this research is to improv e both active and passive approaches to respiratory motion management. The first aim is an improvement to active motion management which focuses on respiratory gating. The second aim focus es on reducing BOT with the use of FFF beams as an i mprovement to passive approaches to respiratory motion management . Specific aims of the research are discussed in the following sections. Aim 1: To develop a respiratory motion management system based on an MR and CT compatible noninvasive respiration sensor for use i n radiotherapy The need for a respiration sensor that is an improvement to currently used commercially available sensors prompted the development of a new respiration sensor to be used with respiratory gating for radiotherapy. It was hypothesized that a t emperature based optical fiber sensor could be used for this purpose . A temperature based sensor provide s motion monitoring by measuring the temperature of inspiratory and expiratory air to track the breathing cycle. The selection of sensor type was base d on the need for a sensor that is compatible in ionizing radiation and MR environments, and would not be subject to the previously discussed li mitations of an external marker. It is desirable for r espiration sensors to be reproducible, accurate, and have a quick response. Additionally, patient comfort was of great importance. Aim 2: To improve clinical implementation of a new technique currently used for passive respiratory motion management Part 1. Dosimetric evaluation of the i nterplay effect with FFF beams The aim of this component of the research was to evaluate the dose delivery accuracy using a moving phantom to simulate respiratory tumor motion for non gated VMAT treatment of moving targets using FFF beams . The interplay effect has been studi ed for IMRT and VMAT with flattened beams, and it has been shown that dose deviations can be reduced by decreasing the dose rate. This project investigated the impact of the interplay effect when using high FFF dose rate for moving targets . Part 2. Use of FFF beams for off axis targets The purpose of this component of the research was to quantitatively characterize methods for dose planning and delivery for lateral lesions. Parameters associated with implementing FFF treatments in off axis geometry ( OAG ) were
26 systematically investigated in this work . Also, it wa s hypothesized that rotational error will have an increased effect on the dose distribution for OAG compared with central axis geometry ( CAG ) . The effect of collimator and couch mechanical error w as studied with an error uncertainty analy sis to determine whether the delivery uncertainty in OAG is acceptable in terms of meeting the objectives of the treatment plan. The following chapters of this dissertation are dedicated to the discussion of each of these specific aims in detail. Each chapter corresponds to a particular project outlined above and gives background, methods, and results obtained.
27 CHAPTER 2 T EMPERATURE BASED OPTICAL FIBER RESPIRATION SENSOR Background A number of respiration sen sors have been investigated including thermistors, thermocouples, strain gauges, pneumotachograph, and external markers [ 7, 38 ]. Kubo et al. found that the temperature sensors showed optimal results in terms of reproducibility, accuracy, quick response, c omfort, and large signal to noise ratio (SNR ) [ 7 ]. Optic fiber based devices that employ temperature sensing have been shown to be successful as respiratory sensors [ 39 43 ]. The use of a thermochromic pigment based fiber optic sensor was studied to measu re respiratory signals during MR image acquisition [ 39 ]. It was shown that this sensor was suitable for respiratory monitoring inside an MR room. The respiratory motion management system proposed for this work requires a temperature sensor to track the temperature sensor in place. The system developed must hold the sensor in a consistent and reproducible position, protect the fiber optic cable from damage, shield the temperature sensor from changes in ambient temperature, and be comfortable for the patient. The performance of the sensor system must be tested in ionizing radiation therapy environments, including CT simulation and high energy radiation treatment rooms. The sensor must also be tested for compatibility in a n MR environment to be consistent with the future direction of radiation therapy technology. Materials and Methods The respiration sensor developed for this work was designed to be an improvement to currently available respiration sensor s for radiotherapy. A temperature
28 based respiration sensor was chosen because it is noninvasive, comfortable, and allows normal breathing so that it is well tolerated by the patient. It does not rely on abdominal or chest movement like external fiducial markers. Plastic fiber optic temperature sensors were investigated because they have been shown to have high transmission rates for fast response times and are feasible in MR environments [ 39, 41 ]. Further b enefits of p lastic optical fibers include good flexibility, long length, small size, low weight, resistance to harsh environments , and no interference with electromagnetic fields. The selection criteria for the fiber optic temperature sensor included temperature range, size , and response time. The t ypical temperature range of inspiratory and expiratory air was measured to be between 22 and 35 degrees Celsius. S ensor s ize was important because s maller sensors have lower heat capacity and therefore better response time. It was desirable to have a res ponse time similar to the RPM system which has been shown to be approximately 0.1 second [ 44 ]. Three candidates were tested before selecting a temperature senso r : STF Flexible Fiber Optic Thermometer for R&D Environments , STB Flexible Fiber Optic Thermom eter for R&D Environments , and MicroProbe Flexible Fiber Optic Thermometer for R&D Environments (LumaSense Technologies, Santa Clara, California). As part of this work the response time was measured for the different temperature probes. The MicroProbe wa s selected because it had the fastest measured response time (0.2 s). It has the smallest diameter (0.12 mm versus 0.80 mm for the STF probe and 0.50 mm for the STB probe ) and a temperature range of 0 120 degrees Celsius. Also , because it is thinner than the other two it performs better in air.
29 Materials and Experimental Setup A respiratory cycle tracking system was fabricated for use in respiratory gating for radiation therapy. The MicroProbe Flexible Fiber Optic Thermometer was used to measure the temp erature of inspiratory and expiratory air to track the breathing cycle. The MicroProbe uses fiber optic thermometry, which uses a phosphor as the sensor. The phosphor radiates ultraviolet light, and the rate of decay of the aft erglow varies with temperat ure. A respiratory system was developed consisting of a mask that can be attached to a plastic tube to provide comfort for the patient and protect the fiber optic cable ( Figure 2 1 and Figure 2 2 ). This system produces a consistent and reproducible posit ion of the sensor and shields the sensing mechanism from changes in ambient air temperature . It also allows for free breathing through either the nose or mouth. Respiratory Cycle Tracking with Volunteer Subjects The respiration signal was initially tested with 3 healthy volunteers for feasibility. Each volunteer was given a mask to wear for the test and instructed to breathe normally. The temperature sensor system reco rded the breathing cycle for the volunteers and the mask system was well tolerated. Fo r the next phase , the respiration signal was tested with an additional 5 healthy volunteers using FSB with compression and compared to simultaneously recorded respiration signal from the RPM system. FSB was employed using an in house compression belt that consists of a blood pressure bladder and Velcro strap (Figure 2 3) . The RPM external marker was placed on the volunteer abdomen , and the signal from the RPM system was recorded simultaneously with the signal from the temperature sensor.
30 CT, Linear Acc elerator, and MR Compatibility The temperature sensor performance was tested in CT, linear accelerator, and MR environments. For the CT testing t emperature readings were acquired by the fiber optic sensor during a typical CT scan for lung patients (GE Lig ht Speed RT CT Scanner; GE Healthcare, Waukesha, Wisconsin). The ambient room temperature was recorded with a calibrated mercury thermometer, and used to calibrate the fiber optic temperature sensor. Background temperature was measured for the fiber optic sensor while the CT was not scanning. Measurements were made during scanning with the fiber optic sensor at the center of the CT field of view and 10 cm outside the field. Measured temperature and e nvironmental induced fluctuations in the temperature se nsor signal were compared to background si gnal . To test the sensor in a linear acc elerator radiation environment, temperature and environmenta lly induced signal variation were measured with the fiber optic cable placed in the direct path of the beam (Var ian TrueBeam; Varian Medical Systems, Palo Alto, California). The sensor was at 90 cm source to surface distance (SSD) with a 25 x 25 cm field size. 100 MU were delivered for 6, 10, and 15 MV beams at 600 MU/min (the maximum dose rate available for FF be ams) . FFF beams were also delivered using the maximum dose rate available for FFF : 1400 MU/min for 6FFF and 2400 MU/min for 10 FFF. Measurements were made for FFF beams with both 100 MU and 3000 MU. During treatment the fiber optic cable would not need t o be in the radiation field, so measurements were also made with the sensor out of the collimated radiation field to simulate the clinical situation. For these measurements the sensor was at 90 cm SSD, and placed 5 cm from the edge of a 10 x 10 field . 30 00 MU were delivered for
31 6 FFF and 10 FFF using the maximum dose rate for each (1400 MU/min and 2400 MU/min, respectively) . The respiration signal from the fiber optic sensor was evaluated in an MR environment with a healthy volunteer. The volunteer was s etup on the MR scanner table and the respiration sensor tracked the breathing cycle of the volunteer while no MR scanning was being done. MR scanning was then initiated and respiration signal was recorded for comparison. quality was investigated. Phantom images were acquired with an MR (Siemens Magnetom 3T Skyra; Siemens Corporation, Washington, D.C.) using the American College of Radiology (ACR) pha n tom and scan protocol for quality assurance (QA) [ 45 ]. The temperature sensor was then secured to the phantom and the scans were repeated. ACR QA tests were performed for each set of scans . The ACR QA tests analyzed were high contrast spatial resolution, low contrast object detectability, geometric accuracy, image intensity uniformity, SNR, and an evaluation of artifacts. Image intensity was measured using the percentage integral uniformity (PIU), which is defined as where h igh and low refer to high and low signal values in the MR image. ACR QA test s for CT were also performed to investigate if the temperature sensor had an effect on CT image quality . First, a CT scan was done with only the QA phantom on the table. Then the CT scan was repeated with the sensor secured to the surface of the phantom and turned on. ACR QA tests evaluated were contrast scale, high contrast spatial resolution, low contrast object detectability, uniformity , and an
32 evaluation of artifacts . Image quality QA tests were analyzed for each scan and compared. Results Test of Respiratory Cycle Tracking with Volunteer Subjects The fiber optic temperature sensor system recorded the breathing cycle for all five volunteers. The positions of the peaks and valleys in the volunteer breathing phases measured by the RPM system matched those of the temperature sensor within an average of 0.16 second. The RPM system requires adequate displacement of the external marker to track the breathing cycle. For one of the volunteers there was not enough displacement and the motion data collecte d from the RPM system was not able to be used for a 4 D CT study. Sample data for two of the v olunteers are shown in Figures 2 4 and 2 5 . Test of CT, Linear Accelerator, and MR Compatibility The fiber optic temperature sensor signal and environmental in duced fluctuations for the CT radiation test are shown in Table 2 1. The average temperature measured by the fiber optic sensor during scanning was less than 0.10 degree Celsius from the average background temperature (22.89 degrees Celsius) measured by t he fiber optic sensor. Standard deviation in temperature signal was also simi lar to background. The fiber optic temperature sensor signal in the linear accelerator high radiation environment is shown in Table 2 2 . The largest difference from the average background temperature during beam on was measured with the senso r in the radiation field for 10 FFF and 3000 MU. This difference was 0.35 degrees Celsius. The standard deviation of the tempe rature signal for 10 FFF and 3000 MU was 0.45 degrees Celsius, w hich was larger than that recorded for background (0.30 degrees Celsius). However,
33 when the sensor was placed 5 cm from the edge of the radiation field fluctuations decreased (standard deviation of 0.24 degrees Celsius) and the difference from average bac kground temperature was 0.11 degrees Celsius. The volunteer breathing cycles with and without MR image acquisition are shown in Figure 2 6 . No significant difference was seen in the amplitude of the breathing cycles ( Table 2 3 ). There was a slight diffe rence in period that is attributed to actual difference in the volunteer breathing cycle since the breathing cycles were not recorded at the same time. MR scans with and without the fiber optic temperature sensor passed all MR imaging QA tests as accepted by the ACR . Results of the ACR QA tests are shown in Table 2 4 . For high contrast spatial resolution, the smallest hole array (0.9 mm) could be resolved in both directions on both exams. All 10 spokes of low contrast objects were visualized on both exam s ( Figure 2 7 ). Both scans had similar results for geometric accuracy, PIU and SNR. Small variations between the results with and without the sensor fall within the daily variability of QA tests for this scanner. No artifacts were visible that could be attributed to the temperature sensor. Both scans with and without the temperature sensor passed all daily CT QA tests as accepted by the ACR. Results of the ACR QA tests are shown in Table 2 5. Slight differences in results of the QA tests are within the range of day to day repeatability variation for that CT scanner. No artifacts were visible that could be attributed to the temperature sensor. Discussion The aim of this component of the research was to develop a respiration sensor to be used with respi ratory gating for radiotherapy that is an improvement to current
34 commercially available sensors. A non invasive temperature based system was developed that can track patient breathing cycles for radiation therapy treatment. The mask system developed is c omfortable for patient use, disposable, and protects the sensor from environmental conditions and changes in surrounding temperature. The mask system was well tolerated by all volunteer subjects. The fiber optic temperature sensor tracked the breathing c ycles of all volunteers, and the breathing phase matched that measured with the RPM system. The motivation for this work was the 11.5% of respiratory gating patients at Mayo Clinic Florida who experienced RPM gating failures caused by having a gating mot ion range less than the required 3 mm. A gating failure was s een by the RPM system during the volunteer testing because there was not enough movement of the external marker for the breathing signal to be usable for a 4D CT. Shallow breathing has little i mpact on the measured respiratory signal of the temperature sensor and it is anticipated to be usable for 4D CT acquisition. Measured temperature and environmental induced fluctuations in temperature signal were similar to background in both a CT and MR en vironment. The largest signal fluctuations occurred when t he sensor was located in the 10 FFF radiation field from a high energy linear accelerator and exposed to a high dose of 3000 MU. Even the largest fluctuations were small compared to the amplitude o f a normal breathing cycle so the temperature sensor signal is still usable to track the breathing cycle. When the sensor was outside the radiation field for the same number of MU the fluctuations were the same as background fluctuations. It is not neces sary for the fiber optic sensor to be
35 in the treatment field, and it is recommended that the sensor be placed outside the treatment field during treatment. Image quality was not affected by the temperature sensor for CT or MR. The results for QA tests w ith and without the fiber optic sensor are within the daily variability of QA tests for each scanner as evidenced by the year to date results for each test. For example, the MR SNR with and without the fiber optic sensor was 512.8 and 504.3, respectively, which was within the range of daily results for that scanner (486.1 to 594.2 with a standard deviation of 22.5). The MR PIU with and without the fiber optic sensor was 92.4% and 91.9%, respectively, which was within the range of daily results for that sc anner (91.4% to 93.2% with a standard deviation of 0.4%). The temperature based respiratory tracking system is inexpensive and reasonably durable. Calibration is done at the time of first use and does not need to be repeated for daily use. Additionally t here is no warm up time so use of the tracking system would not cause any delays, which is an essential requirement in the clinic. In conclusion, a new respiration sensor was developed that tracks the breathing cycle by measuring the temperature of inspirato ry and expiratory air . The temperature based optical fiber respiration sensor that was developed tracked the breathing cycles of all volunteers even when the external marker failed. The mask system was well tolerated by all volunteers. The mask system p uts the sensor in a reproducible position and protects the fiber optic cable. The temperature based sensor maintained performance in MR and ionizing radiation environments. It caused no deterioration of CT or MR images. Results show that the device is s uitable for respiratory gating and an improvement to external fiducial markers.
36 Table 2 1 . Fiber optic temperature sensor signal during CT scanning. Note: Background readings were made while the CT was not scanning. Table 2 2. Fib er optic temperature sensor signal in high energy linear accelerator environment. Note: Background readings were made while the linear accelerator radiation beam was not on. Standard Deviation Difference from background temperature (degrees Celsius) Standard Deviation (degree s Celsius) Background Temperature sensor in radiation field 0.36 0.02 0.30 Temperature sensor 10 cm from radiation field 0.07 0.31 Standard Deviation (degrees Celsius) Energy (MV) Monitor Units Difference from background temperature (degrees Celsius) Standard Deviation (degrees Celsius) B ackground Temperature sensor in radiation field 0.3 0 6 100 0.13 0.14 10 100 0.11 0.17 15 100 0.14 0.14 6 FFF 100 0.04 0.10 10 FFF 100 0.09 0.20 6 FFF 3000 0.07 0.29 10 FFF 3000 0.35 0.45 Temperature sensor 5 cm from edge of radiati on field 6 FFF 3000 0.13 0.24 10 FFF 3000 0.11 0.24
37 Table 2 3. Respiration signal recorded by fiber optic temperature sensor in MR environment. T emperature ( degrees Celsius ) (standard d eviation) MRI not scanning MRI scanning Average Max 32.7 (0.1) 32.7 (0.1) Average Min 23.3 (0.3) 23.7 (0.1) Average Amplitude 9.4 (0.4) 9.0 (0.1) Time (seconds) (standard d eviation) Average Period 5.5 (0.0) 5.9 (0.4) Table 2 4. MR American College of Radiology Quality Assurance. Table 2 5. CT American College of Radiology Quality Assurance. Exam High Contrast Resolution Low Contrast Detectability X Axis Length (mm) Y Axis Length (mm) Z Axis Length (mm) ACR SNR PIU With Temperature Sensor 0.9/0. 9 10 190.43 191.41 149.41 512.8 92.4 Without Temperature Sensor 0.9/0.9 10 190.43 191.41 148.93 504.3 91.9 Exam Contrast Scale (mean CT#) High Contrast Spatial Resolution (standard deviation) Low Contrast Detectability (# of objects seen) Uni formity (Central mean CT# Edge mean CT#) With Temperature Sensor 122.84 38.43 4 0.87 Without Temperature Sensor 122.32 38.58 4 1.19
38 Figure 2 1. Fiber optic temperature senso r in plastic holder . Photo courtesy of author. Figure 2 2. Mask system consists of plastic holder attached to disposable mask . Photo courtesy of author.
39 Figure 2 3 . Temperature based respiration system and RPM external marker used to simultaneousl y record volunteer breathing cycle with FSB . Photo courtesy of author.
40 Figure 2 4. Sample breathing data (both sensors tracked cycle). Figure 2 5. Sample breathing data ( RPM data not usable for 4D CT).
41 Figure 2 6. Volunteer b reathing cy cles in MR environment.
42 Figure 2 7. MR low contrast object detectability with no sensor (left) and with sensor (right).
43 CHAPTER 3 DOSIMETRIC EVALUATION OF THE INTERPLAY EFFECT WITH FFF BEAMS Background VMAT has been widely implement ed clinically as a part of SBRT for treatmen t of lung and liver cancer [ 46, 47 ]. This delivery technique provides high dosimetric conformity which is required in SBRT techniques. VMAT has similar dosimetric quality to IMRT with a more efficient delivery. However, in both IMRT and VMAT, interplay effects exist when treating targets that are influenced by respiratory motion . The interplay effect is a combination of tumor motion caused by respiration and beam motion (movement of the dynamic MLC as it shape s the field to modulate the dose). This effect is only an issue for dynamic delivery techniques, such as IMRT and VMAT because with these techniques beam intensity gradients are no longer confined solely to the edges of the beams. Gradients can be inside the field defined by the primary collimators. If a target is moving inside this field dosimetric error can be introduced. The interplay effect has been studied for radiation beams that u se a FF [ 48 53 ]. Variation in dose to moving targets has been fou nd to be most significant for individual fields and single fractions. Dose deviations in IMRT plans can be decrea sed by reducing the dose rate [ 48 , 52 ], but it is questionable whether dose rate will impact VMAT dose deviations. This is due to the complic ated delivery technique that VMAT uses to accomplish the desired dose distribution, including modulating the dose rate, gantry rotation speed, and MLC leaf moving speed. It has been found that FFF mode gives similar dose distributions compa red to FF beams for lung SBRT with reduced BOT [ 54 ]. VMAT combined with FFF mode provides an opportunity for higher fraction doses to be delivered with high dosimetric
44 performance and efficiency. FFF beams have a faster BOT by using a significantly higher dose rate than FF beams. For clinical implementation of FFF beams with VMAT it is important to investigate the impact of FFF mode on the interplay effect , which is done in this work . Because it has been shown previously that reducing the dose rate will decrease motion induced dose deviations, the dose delivery accuracy of treating moving targets is investigated with the significantly higher dose rates associated with FFF mode. A recent study showed the interplay effect had limited impact on VMAT treatments using FFF be ams when res piratory gating was utilized [55 ]. Non gated VMAT treatments need to be evaluated as well because respiratory gating may not be used during treatment for a number of reasons. First, not all clinical sites have the hardware required for respir atory gating. Second, patients may not be a candidate for respiratory gated treatments due to irregular or shallow breathing cycles, or if the gating signal is compromised due to the FSB technique or immobilization. Third, gated treatments may not be use d if the prolonged treatment time associated with respiratory gating is impractical for the patient. It has been found that the average treatment time was about three times longer for VMAT with gating compared to non gated VMAT delivery [ 56 ]. The aim of t his component of the research was to evaluate dose delivery accuracy using a moving phantom to simulate respiratory tumor motion for non gated VMAT treatment of moving targets using FFF beams, and to investigate the impact of high FFF dose rate on the inte rplay effect.
45 Materials and Methods Two patient plans were selected from lu ng SBRT patients treated with 6 FFF on a Varian TrueBeam LINAC having a tumor size less than 5 cm. FSB and the motion encompassing method were used for motion management. Respira tory gating was utilized during CT simulation to create a 4D CT set to produce a tumor motion encompassing volume. Two arcs were used for each patient plan; one plan had two full arc rotations of the gan try and one had two partial arc rotations of the gan try in order to avoid collision of the gantry head with the patient, treatment couch , or treatment devices . The plans were copied to a phantom and recalculated on the Pinnacle treatment planning system (Pinnacle Version 9 software; Philips Medical Systems , N.A., Bothell, Washington). The planned dose was delivered with the Varian TrueBeam. Dose distributions were measured using a 2D array consisting of 1020 ionization chambers (ImRT MatriXX; IBA, Bartlett, TN). The detector array was placed on top of a m otor driven motion platform (Respiratory Gating Platform; Standard Imaging, Middleton, WI) to simulate patient breathing ( Figure 3 1 ). The motion of the platform wa s in one dimension in the superior inferior direction with an amplitude of 30 mm and a perio d of 4 seconds. In order to compare the delivery for three different dose rates for each patient the breathing cycle phase was taken into account. The phantom motion was divided into 8 equally spaced phases and irradiation was initiated at each of the pha ses. The 8 initial starting phases were used to simulate the clinical situation of starting the treatment at a random starting phase in the patient breathing cycle. A video monitoring system was used to visually determine when the phantom was in a partic ular phase. This visual method of initiating treatment has been used by others, and has been shown to
46 introduce negligible uncertainty [ 48, 49 ]. Delivery of each plan was done for all fields in their treatment geometries with and without phantom motion u sing 3 dose rates: 1400, 600, and 400 MU/min. For each phase the delivery was compared to the treatment plan using gamma analysis (pass criteria of 3%/3mm) and absolute difference (percent of pixels within 5% and 10% of planned pixel dose) using OmniPro so ftware (OmniPro ImRT; IBA, Bartlett, TN). Gamma analysis qualitatively compares dose distributions using dose and distance criteria (3% and 3 mm in this work). Gamma is calculated independently for each reference point and identifies regions where the dos e difference and distance to agreement are simultaneously greater than the criteria. Calculations were done for each of the 8 initial starting phases. In order to systematically account for the situation of initiating treatment at a random point in the p atient breathing cycle, the calculations were then repeated for 1000 trials of randomly selected initial starting phases. Next, dose difference was investigated for 4 points at the center of the target to evaluate dose to the gross tumor volume. The poin ts were chosen to coincide with the 4 ionization chambers closest to the central axis in the ionization chamber array. Four peripheral points were also selected at the edges of the target. Again these points coincided with ionization chambers in the arra y. An example of the peripheral points for Plan 1 is shown in Figure 3 2 . Results The results for the static and moving delivery are reported for 1000 trials of randomly selected starting phases for gamma analy sis (3%/3mm) and absolute difference ( percen t of pixels within 5% and 10% of planned pixel dose) in Table 3 1 . For both plans more than 95% of the measured pixels were within 3%/3mm for the
47 static condition, whereas that dropped to approximately 86% for the moving condition for all dose rates . The percent of pixels within 5% of planned pixel dose was approximately 93%, which dropped to approximately 86% for the moving condition for all dose rates . Both static plans had greater than 97% of pixels within 10% of the planned pixel dose, whereas that d ropped to approximately 94% for the moving condition for all dose rates . The delivered dose compared to planned dose for 4 points at the center of the target and 4 points at the periphery of the target is shown in Table 3 2 . The dose difference in the cen ter of the target region was within 5 .0 % for the static phantom condition and within 6.9 % for the moving phantom condition. At the periphery of the target the delivered dose difference was within 13 .0 % for the static condition and 38.5 % for the moving con dition. Discussion The goal of this component of the research was to evaluate dose delivery accuracy using a moving phantom to simulate respiratory tumor motion for non gated VMAT treatment of moving targets using FFF beams, and to investigate the impact o f high FFF dose rate on the interplay effect. The impact of the interplay effect was seen for all plan quality parameters analyzed as expected . No substantial differences were seen between different dose rates in the moving phantom study. The largest di fference in gamma values between different dose rates was less than 0.5%. In a previous study VMAT deliveries were found to be more tolerant to variations in gantry position and MLC leaf posi tion than IMRT [ 57 ]. The changing dose rate during VMAT deliver y that is used to accomplish modulation may reduce the impact of dose rate on dose differences.
48 There was a slight degradation of dose distribution in the target region when the phantom was moving (5.0% vs 6.9%). The dose delivered to the moving target wa s always greater than the planned dose, so the target was never under dosed. Dose difference at the periphery was greater in magnitude, which was expected due to the interplay effect in a high gradient region. A recent dosimetric study of the interplay e ffect with FFF beams found that FFF dose distributions were comparable to FF beams as long as a sufficient margin was given [ 58 ]. Our results agree that target dose is acceptable for FFF with moving targets as evidenced by the dose delivered to central poi nts. A treatment margin of up to 1.0 cm will ensure acceptable target coverage; however selection of margins must be balanced with increased dose to surrounding structures and is a clinical decision.
49 Table 3 1 . Delivered vs planned dose distribution f or 1000 tria ls of randomly selected starting phases . Phantom Dose Rate (MU/min) Gamma 3%/3mm Standard deviation % of pixels within 5% of dose Standard deviation % of pixels within 10% of dose Standard deviation Static 95.51 92.53 97.06 Plan 1 Moving 1400 86.93 0.84 86.49 0.25 94.06 0.13 600 86.77 0.23 86.55 0.07 94.08 0.04 400 86.64 0.29 86.57 0.04 94.08 0.02 Static 97.95 93.80 97.70 Plan 2 Moving 1400 85.86 0.24 85.88 0.24 93.94 0.24 600 86.22 0 .36 85.78 0.06 93.89 0.04 400 86.30 0.08 85.84 0.07 93.89 0.05 Table 3 2 . Dose difference for delivered v s planned dose for central and peripheral target points . Moving Static Dose Rate (MU/min) Points 1400 600 400 1 5.5% 5.2% 5 .0% 5.0% 2 3.2% 3.1% 2.9% 2.2% center 3 2.8% 2.5% 2.6% 1.9% Plan 1 4 6.8% 6.9% 6.5% 3.4% 1 34.6% 34.4% 34.7% 13.0% 2 3.0% 2.7% 2.4% 5.7% periphery 3 28.3% 28.4% 28.4% 10.1% 4 17.3% 17.2% 16.6% 0.4% 1 3.9% 4.9% 4.2% 0.6% 2 1.0% 0.3% 0.3% 1.2% center 3 1.8% 2.3% 2.1% 1.6% Plan 2 4 2.4% 3.1% 2.3% 0.4% 1 14.3% 14.2% 14.0% 7.6% 2 23.4% 23.4% 22.7% 5.9% periphery 3 16.9% 15.8% 15.7% 0.9% 4 37.7% 38.5% 38.1% 5.9%
50 Figure 3 1. 2D ion ization chamber array on motor driven motion platform (Respiratory Gating Platform; Standard Imaging, Middleton, WI). Photo courtesy of author.
51 Figure 3 2. Points selected at the periphery of the target region (OmniPro ImRT; IBA, Bartlett, TN).
52 CHAPTER 4 USE OF FFF BEAMS FOR OFF AXIS TARGETS Background DCAT in full co planar rotation is commonly adopted for efficient conformal dose delivery in SBRT [ 5 9, 60 ]. SBRT can benefit from the use of FFF beams because SBRT uses a high dose per fraction. The increased efficiency of FFF beams also aid s i n respiratory motion management for SBRT when using DCAT by decreasing the potential for baseline shifts and a llowing for the use of breath hold techniques [29, 30 ]. For laterally located targets, which are commonly seen in lung and liver SBRT, it is often necessary to use OAG to obtain full rotation of the gantry without collision of the gantry head with either t he patient , the treatment couch , or treatment devices . OAG occurs when the isocenter is positioned at patient midline; however, the target is loca ted laterally off axis (Figure 4 1 ). CAG occurs when the isocenter is located at the center of the target, reg ardless of its position within the patient. However, when the target is located too far laterally, a full arc rotation is not possible with CAG because the gantry head which rotates around the patient will collide with the patient, treatment couch, or tre atment devices. Not having a full arc rotation of the gantry could have a negative impact on the dose distribution. The use of a modified DCAT technique, which utilizes a midline isocenter, has been shown to be useful for the treatment of peripheral lung tumors using SBRT for FF beams [ 61, 62 ] . These studies have demonstrated that the use of a modified DCAT technique results in enhanced planning target volume ( PTV ) coverage, improved conformality, reduced BOT, and meets the requirements of Radiation Therapy Oncology Group ( RTOG ) protocols 0915 and 0813 for lung SBRT [ 63, 64 ] . However, this technique has only been investigated using FF beams. Due to
53 the significantly non un iform incident fluence profile of FFF bea ms, i t is important to study this technique with FFF beams, which was done in this work. While the incident fluence profile of FFF beams is significantly non uniform, there are several advantages of FFF beams compared to conventional beams that could in fluence dosimetry for lateral lesions. FFF beams have less variation of off axis beam hardening due to the removal of the FF from the path of the beam. FFF also has less photon head scatter due to removing the FF , which results in less field size dependenc e, as well as less leakage outside of beam collimation. The peripheral dose far from the field edge (15 20 cm) is influenced by the treatment head leakage, which has been shown to be reduced by 52% for 6 MV and 65% for 10 MV unflattened beams [ 65 ]. The p enumbra size and MLC leakage are also reduced for FFF beams [ 66, 67 ]. All of these factors act to improve the software modeling accuracy of photon beams. While these advantages make it desirable to use FFF beams, it is necessary to quantitatively charact erize methods for dose planning and delivery for lateral lesions. Parameters associated with implementing FFF treatments were systematically investigated in this work . Additionally, due to the distance between the axis of rotation and the target for OAG, it is hypothesized that mechanical rotational error will have an increased effect on the dose distribution compared with CAG. The effect of collimator and couch mechanical error was studied with an error uncertainty analy sis to determine whether the delive ry uncertainty in OAG is acceptable in terms of meeting the objectives of the treatment plan. The quantitative treatment plan objectives investigated were surface target dose coverage, conformality indices, and dose to the lung.
54 Materials and Methods Late ral Target, Off Axis Geometry Evaluation A representative CT image set was selected from lung S BRT patients (GE LightSpeed RT CT Scanner; GE Healthcare, Waukesha, Wisconsin). The CT image set was transferred to a Pinnacle treatment planning system (Pinnac le Version 9 software; Philips Medical Systems, N.A., Bothell, Washington). Nine sphere targets having 3 different sizes (2, 4, and 6 cm diameter) were placed at 3 different locations (center of the patient, 3 cm later al, and 6 cm lateral) . The isocenter w as located at the center of the patient for all targets. For each target, DCAT plans were obtained using 4 different beams (6X, 6FFF, 10X, and 10FFF) for a Varian TrueBeam (Varian TrueBeam; Varian Medical Systems, Palo Alto, California). Plans were normali zed such that 95% of the target volume received the prescription dose . For all plans, two conformality indices (CI) were calculated using the RTOG definition ( CI 100 is the prescription isodose volume to target volume and CI 50 is the 50% prescription isod ose volume to target volume) [ 63, 64 ] . The minimum possible CI 100 value is 0.95 with exact 95% coverage. H omogeneity in dex ( HI ) was also calculated for each plan using the definition commonly used for radiosurgery ( ratio of ma x dose to prescription dose) [ 68, 69 ] . BOT was calculated based on planned MU and the maximum dose rate available in each beam mode. The ratio of BOT for FFF to flattened beams was calculated for each target. Mechanical Rotati on Induced Dosimetric Uncertainty Evaluation The effect of collimator and couch rotational uncertainty on plan quality in OAG was evaluated. Five representative patient CT sets were selected from lung SBRT patients . The selection criteria for the patient CT sets was tumor size less than 163 cc and tumors located 5 cm or more from patient midline. The tumor volumes were 24.4
55 cc, 28.7 cc, 113.1 cc, 57.9 cc, and 107.5 cc. The tumors ranged from 5.1 to 9.7 cm in lateral distance from the midline of the patie nt to the center of the tumor. Each case was retrospectively planned with Pinnacle software using 6FFF fo r a Varian TrueBeam. For each of the 5 cas es , a plan with a single full arc gantry rotation was generated for OAG with 95% target volume in prescriptio n dose (48 Gy in 4 fractions). The plans met all of the dose objectives of RTOG lung SBRT protocols 0915 and 0813 [ 63, 64 ] . This original plan was then copied and 3 new plans were generated with the following modifications: collimator rotation of 1.0 degr ee, couch rotation of 1.0 degree, and both collimator and couch rotations of 1.0 degree. One degree was chosen as a conservatively large estimate of rotation errors. The same MU settings were used for all plans. The same procedure was then repeated for CAG with the same prescription dose and normalization; however, only a partial arc gantry rotation was used to avoid gantry collision . A full 360 degree arc rotation would not be deliverable with CAG because of a collision of the gantry with either the patie nt or treatment couch due to the lateral position of the isocenter. A total of 8 plans were generated for each of the 5 cases. Each plan was evaluated for PTV coverage, dose conformality, and critical organ dose. PTV coverage was evaluated by the prescript ion isodose surface coverage (percent of the target volume receiving the prescription dose) for each of the plans. Conformality indices investigated were CI 100 and CI 50. The V20 of lung (percent of total normal lung receiving 20 Gy or more) was also dete rmined per RTOG protocol [ 63, 64 ] . Statistical A nalysis For the mechanical uncertainty evaluation, absolute differences were calculated (i.e. , the absolute value of the differ ence) from the original plan for 1.0 degree collimator
56 rotation, 1.0 degree couch rotation, and both collimator and couch rotations of 1.0 degree. These absolute differences were calculated separately for OAG and CAG for: prescription isodose surface coverage, CI 100, CI 50, and lung V20. The absolute differences from the original pla n were compared between OAG and CAG using a paired t test. P values of 0.05 or lower were considered statistically significant. Statistical analyses were performed using R Statistical Software (version 2.14.0; R Foundation for Statistical Computing, Vienna , Austria). Results Lateral Target, Off Axis Geometry Evaluation Results for the evaluation of usage of FFF beams for off axis, lateral targets compared to central axis targets for 6 MV and 10 MV photon beams are shown in Table 4 1 and Table 4 2 , respect ively . 6FFF HI values were similar to 6X for all targets. 10FFF showed consistently larger HI values than 10X for all targets. Though differences in HI values were larger for 10 MV beams than for 6 MV beams, 10 MV did not show considerable difference bet ween flattened and unflattened beams, with the maximum HI difference being 0.10 (3 cm off axis, 6 cm diameter target) . 6FFF CI 100 values were similar to 6X for both CAG and OAG for all target sizes. In contrast, CI 100 values were similar for 10 MV flat tened and unflattened beams in CAG, but for targets 3 cm OAG and 6 cm OAG, 10FFF C I 100 values were slightly higher compared to 10X. FFF beams gave lower CI 50 values for 2 cm targets for 6 and 10 MV in CAG and OAG. The advantage of FFF for CI 50 declined with increasing target diameter. Using FFF beams always reduced the BOT, as evidenced by ratios of BOT from 0.26 0.32 for 10 MV and 0.46 0.51 for 6 MV. The benefit of reduced BOT with FFF decreased with off axis distance of the target. The use of FFF be ams with OAG always
57 increased the MU. This effect was greatest for the large size targets at 6 cm off axis. The maximum MU increase for FFF to flattened beams was 19% for 6 MV and 28% for 10 MV. Mechanical Rotation Induced Dosimetric Uncertainty Evalua tion The isodose coverage, conformality, and V20 for all plans with mechanical rotation induced dosimetric uncertainty met the requirements for RTOG protocols. Figure 4 2 shows a representative axial slice for each case and results for the uncertainty eval uation are found in Tables 4 3 and 4 4 . As displayed in Table 4 4 , the absolute difference in prescription isodose surface coverage from the original plan for collimator rotation, couch rotation, and collimator plus couch rotation s of 1.0 degree ranged fr om 0.07% to 3.02% for OAG and from 0.00% to 0.24 % for CAG. In general, the effect of simulated couch or collimator rotation was to decrease the prescription isodose coverage with the largest difference being 3.02% (collimator plus couch rotation for OAG). P values of 0.05 or lower were considered statistically significant . The variation from the original plan was significantly smaller for CAG compared to OAG for collimator rotation (P=0.02 ) and collima tor plus couch rotation (P=0.03 ), with a similar, thoug h not quite significant fin ding for couch rotation (P=0.08 ). Differences in prescription isodose surface coverage for each plan are further summarized in Figure 4 3 . The conformality was also evaluated for each plan. The absolute difference in CI 100 a nd CI 50 from original OAG and CAG plans for mechanical rotational error is shown in Table 4 4 . These absolute differences from baseline regarding collimator rotation, couch rotation, and collimator plus couch rotations were generally similar for CAG comp ared ), with one small, but statistically significant difference
58 occurring regarding C I 50 for couch rotation (P=0.03 ). The conformality was found to increase or decrease depending on the shape of the target (Figure 4 2 ). CI 100 and CI 50 for OAG and CAG for collimator and couch rotational error are further illustrated in Figures 4 4 and 4 5 . The V20 for normal lung was calculated for each plan. The mechanical rotation uncertainty improved the V20 in some cases, but caused it to be worse for o ther cases. The V20 was acceptable for all cases with all plans including those with simulated rotational error having a V20 less than 6%. The absolute difference in V20 from the original OAG and CAG plans for collimator rotation, couch rotation, and colli mator plus couch rotations are shown in Table 4 4 and Figure 4 6 . The variation from the original plan was relatively small for all cases; although it was smaller for CAG compared to OAG for collima tor plus couch rotation (P=0.03 ). Discussion FFF plans w ere evaluated for lateral targets with homogeneity and conformality indices to compare CAG and OAG planning techniques. The results of this study demonstrate that FFF beams can be utilized in OAG without dosimetric compromise. Differences in HI values bet ween FFF and FF beams for both 6 and 10 MV were small. For both 6 and 10 MV beams, FFF provided a benefit for CI 50 for small targets in CAG and OAG. The advantage of FFF for CI 50 declined with target diameter. 10FFF can provide more dramatic BOT reducti on compared to 6FFF; however, the benefit decreases with target off axis distance when treating with OAG. Additionally, the effect of collimator and couch mechanical error was studied with an error uncertainty analy sis. D ue to the distance between the axis of rotation and the target for OAG, it was hypothesized that rotational error would have an increased effect
59 on the dose distribution compared with CAG. The mechanical rotation induc ed dosimetric variation was greater for OAG compared to C AG, with so me indices showing statistical sig nificance, as seen in Table 4 4 . However , the differences between original plans and plan s with simulated mechanical rotation were small in both CAG and OAG, and will not have a clinical impact . The target isodose coverage decreased by an average of 1.00% for OAG when rotation error was present and by an average of 0.1 0% for CAG. The maximum absolute difference caused by rotational error was 0.05 % and 0.08 % for conformality indices and lung V20, respectively. All lung V20 r esults in this study including those with simulated rotational error were less than 6%, which me e t s the RTOG dose objectives of less than 10%. It should be noted that this study investigated the effects of one fraction. If additional fractions are used and the rotational error is random, then the dosimetric variation would be reduced. The collimator and couch rotation investigated in this study was 1 .0 degree, which is a conservatively large estimate to what may be expected in clinical practice. The tole rance for collimator or couch rotation is 0.5 degree. At our institution, annual quality assurance of the True B eam has demonstrated less than 0.3 degree and 0.2 degree for collimator and couch angle indicators, respectively. Therefore, the dosimetric var iation due to rotation al error found in this study is conservatively large . While this study found that the effect of mechanical rotation error with OAG is acceptable, patient rotation during treatment could also have an effect on the delivered dose distr ibution when treating in OAG. Josipovic et al . investigated intra and interfractional errors in patient position of SBRT and found with correction of translational errors the remaining rotational errors were approximately 1 .0 degree [ 70 ].
60 Use of image gui dance would reduce these patient oriented rotational errors and is important in SBRT. Daily IGRT is necessary to ensure accurate patient alignment for each treatment and is part of the ACR ASTRO practice guidelines for SBRT [ 71 ]. To evaluate which techniqu e gave better dosimetric results, original CAG and OAG plans with no collimator or couch rotation were compared ( Table 4 3 and Figure 4 2 ). This demonstrates OAG gave better dosimetric results for some cases, but not for others. The dosimetric results depe nd on the length of the arc rotation in relation to the tumor shape. For example, OAG with a full arc rotation gave a better dose distribution than CAG for spherical targets. CAG with a partial arc rotation was better than OAG for elliptical targets when the long axis of the target was in the direction of the arc. Both of these effects are more pronounced for low dose, as seen with CI 50. It should be noted that this comparison is influenced by the effect of using a full arc rotation in OAG and a partial arc rotation in CAG. While this is not a fair comparison, it does represent the clinical situation , and may influence the decision of which geometry to use for an individual . This study showed that FFF beams in OAG can provide acceptable dose distributions in terms of HI, CI, and lung V20 . There are benefits and limitations to both OAG and CAG. The selection of which geometry to use depend s on the situation and it may be influenced by whether a full or partial arc is desirable. This study showed an advantag e of using OAG for small, spherical targets, which are often seen in SBRT. Another benefit of OAG is that no table shift is necessary during IGRT where a full gantry rotation is required for cone beam CT. However, it can also be argued that CAG is advantag eous for IGRT because it is desirable to place the isocenter at the location of
61 the target for imaging. Other benefits of CAG include reduced MU and less BOT. In some cases the choice to use OAG or CAG may depend on critical organs. Using a partial arc may reduce critical organ dose. It is also possible that a partial arc may be preferred over a full arc due to the possibility of ret reatment, especially if SBRT is repeated for additional lesions. The results of this study indicate that FFF beams can be used in OAG without dosimetric c ompromise. 10 FFF provides more dramatic BOT reduction but generally less favorable dosimetric indices compared to 6 FFF in OAG. Mechanical uncertainty in collimator and couch rotation had an increased effect for OAG compared to CAG, however the variatio ns in dose distribution for either treatment technique were small and t have a clinical im pact.
62 Table 4 1. Lateral target, off axis geometry evaluation results for 6 MV. Table 4 2. Lateral target, off axis g eometry ev aluation results for 10 MV. Target Central Axis 3 cm Off Axis 6 cm Off Axis Diameter (c m) Flattened FFF Flattened FFF Flattened FFF HI 2 1.19 1.17 1.16 1.15 1.14 1.13 4 1.11 1.11 1.11 1.12 1.12 1.10 6 1.11 1.13 1.11 1.13 1.12 1.11 2 1.11 1.11 1.06 1.06 1.04 1.04 CI 100 4 0.98 0.99 0.99 0.99 1.00 1.00 6 0.98 0.99 0.99 1.0 0 1.00 1.01 2 4.23 4.12 4.33 4.28 4.33 4.30 CI 50 4 3.36 3.36 3.36 3.29 3.39 3.38 6 3.11 3.17 3.17 3.17 3.21 3.24 2 0.46 0.47 0.50 BOT ratio 4 0.47 0.48 0.50 6 0.48 0.49 0.51 Target Central Axis 3 cm Off Axis 6 cm Off Axis Diameter (cm) Flattened FFF Flattened FFF Flattened FFF HI 2 1.23 1.25 1.20 1.21 1.17 1.19 4 1.13 1.17 1.14 1.19 1.14 1.17 6 1.12 1.20 1.13 1.23 1.13 1.21 2 1.12 1.13 1.06 1.08 1.03 1.07 CI 100 4 0.99 0.99 0.99 1.04 0.99 1.03 6 0.98 1.00 0.99 1.04 0.99 1.04 2 4.59 4.49 4.73 4.64 4.71 4.65 CI 50 4 3.48 3.40 3.54 3.53 3.56 3.52 6 3.13 3.12 3.24 3.29 3.25 3.29 2 0.26 0.27 0.3 1 BOT ratio 4 0.27 0.29 0.32 6 0.29 0.30 0.32
63 Table 4 3. Results of prescription isodose surface coverage, conformality indices, and lung V20 for OAG and CAG. Mean (minimum, maxim um) value Variable Original plan 1 degree collimator rotation 1 degree couch rotation 1 degree collimator, 1 degree couch rotation Prescription Isodose Surface Coverage (%) OAG 95.3 3 (95.1 8, 95.56 ) 94.9 1 (94.57 , 95.5 0 ) 93. 9 4 (92.46 , 95.0 0 ) 93. 7 1 (92.3 1 , 94.7 5 ) CAG 95.2 0 (95.1 6 , 95.3 0 ) 95.2 2 (95.2 1 , 95.2 2 ) 95.2 5 (95.0 3 , 95.5 3 ) 95.2 7 (95.0 0 , 95.3 2 ) CI 100 OAG 1.33 (1.20, 1.54) 1.32 (1.19, 1.54) 1.33 (1.19, 1.54) 1.33 (1.19, 1.53) CAG 1.38 (1.19, 1.60) 1. 38 (1.19, 1.60) 1.38 (1.19, 1.60) 1.38 (1.19, 1.60) CI 50 OAG 5.07 (4.05, 5.74) 5.06 (4.05, 5.72) 5.08 (4.04, 5.76) 5.06 (4.05, 5.73) CAG 5.52 (4.75, 7.06) 5.54 (4.76, 7.08) 5.53 (4.74, 7.08) 5.54 (4.75, 7.09) Lung V20 (%) OAG 3.78 (1.22, 5.99) 3.77 (1.21, 5.99) 3.75 (1.20, 5.95) 3.74 (1.19, 5.96) CAG 3.71 (1.19, 6.16) 3.72 (1.19, 6.16) 3.71 (1.18, 6.15) 3.71 (1.18, 6.16)
64 Table 4 4. Absolute difference of mechanical rotation and o riginal plan results for OAG and CAG. Mean (minimum, maximum) value Variable Absolute difference: Collimator Original Absolute difference: Couch Original Absolute difference: Collimator/Couch Original Presc ription Isodose Surface Coverage (%) OAG 0.39 (0.07, 0.76) 1.34 (0.11, 2.87) 1.56 (0.47, 3.02) CAG 0.05 (0.00, 0.09) 0.15 (0.06, 0.24) 0.11 (0.01, 0.21) P=0.02 P=0.08 P=0.03 CI 100 OAG 0.01 (0.00, 0.01) 0.02 (0.00, 0.05) 0.02 (0.01, 0 .04) CAG 0.00 (0.00, 0.01) 0.00 (0.00, 0.01) 0.00 (0.00, 0.02) P=0.18 P=0.21 P=0.21 CI 50 OAG 0.01 (0.00, 0.03) 0.01 (0.00, 0.02) 0.01 (0.00, 0.01) CAG 0.01 (0.00, 0.03) 0.02 (0.01, 0.03) 0.02 (0.00, 0.03) P>0.99 P=0.11 P=0.03 Lung V20 (%) OAG 0.01 (0.00, 0.01) 0.03 (0.01, 0.08) 0.04 (0.02, 0.08) CAG 0.00 (0.00, 0.01) 0.01 (0.00, 0.01) 0.00 (0.00, 0.01) P=0.37 P=0.11 P=0.03 Note: Absolute differences are the absolute values of the given difference. P values are the c omparison of differences from original plan values. P values result from a paired t test.
65 Figure 4 1. CAG and OAG for a laterally located target.
66 Figure 4 2 . Mechanical rotation induced dosimetric uncertainty evaluation for 5 cases. Repre sentative axial slice for OAG (left) and CAG (right).
67 Figure 4 3. Prescription Isodose Surface Coverage for OAG and CAG for the original plan s and mechanical rotation plans . Points and lines for the same patient are shown in the same color.
68 Figure 4 4. CI 100 for OAG and CAG for the original plan s and mechanical rotation plans . Points and lines for the same patient are shown in the same color.
69 Figure 4 5. CI 50 for OAG and CAG for the original plan s and mechanical rotation plans . Points and lin es for the same patient are shown in the same color.
70 Figure 4 6. Lung V20 for OAG and CAG for the original plan s and mechanical rotation plans . Points and lines for the same patient are shown in the same color.
71 CHAPTER 5 CONCLUSIONS Results of This W ork The goal of this work was to improve both active and passive approaches to respiratory motion management for radiation therapy. This is important to the field of medical physics because tumor motion due to respiration is a source of positional uncerta inty of the target volume which requires adequate margins to encompass the tumor resulting in a large volume of irradiated tissue and potentially increased toxicity . There are a number of methods that have been suggested for motion management and each has advantages and disadvantages. It is important to improve both active and passive approaches to motion management because there is not a single motion management technique that suits every situation. The motion management method used for a particular pat ient depends on availability of resources at the facility and whether the patient can tolerate a particular technique. This research contributed to active respiratory motion management techniques with an improvement to respiratory gating . A non in vasive temperature based optical fiber respiration sensor was developed for respiratory gating in radiation therapy. The temperature sensor tracks the breathing cycle by measuring the temperature of inspiratory and expiratory air. A mask system was devel oped that was designed to provide comfort for the patient. The mask system protect s the sensor and puts it in a reproducible position . The mask was well tolerated by all volunteers. The temperature based respiratory gating system tracked the breathing cycles of all volunteers even when gating motion range wa s too small for the commonly used RPM external marker signal to be usable . The fiber optic temperature sensor mainta ins performance in MR
72 and ionizing radiation environment s ; however , it is recommended that it not be used in the direct path of a high energy radiation beam. It has been verified that the respiratory si gnals can be obtained without deterioration of CT or MR images. These results indicate that this device is suitable for respiratory gating in radiotherapy and is an improvement to commonly used external markers. The improvement to passive approaches to mo tion management focused on reducing BOT with the use of FFF beams. Reducing BOT is important in motion management because it aids in the use of breath hold techniques and reduces changes over time in breathing patterns. The use of high dose rate FFF beam s for treating moving targets with VMAT was evaluated. The impact of the interplay effect with FFF beams was observed on dose distributions of non gated VMAT when treating a moving target . H owever , it was shown that using high FFF dose rate did not incre ase dose deviations compared to lower dose rates commonly used with FF beams . FFF beams were also evaluated for treating laterally located targets in OAG , which are often seen in lung and liver SBRT . It was shown that FFF beams can be used in OAG without dosimetric compromise. All plans met the dose objectives of the lung SBRT RTOG 0813 and 0915 protocols and differences between 6FFF and 10FFF were small. 10FFF provid es more BOT reduction than 6FFF; however the benefit of reduced BOT with FFF decrease s with distance of the target off axis for both 6 and 10 MV. Mechanical error in collimator and couch rotation produces increased dosimetric variation for OAG compared to CAG; however, the variations in dose distributions for either treatment technique were small and will not have a clinical impact . These results indicate that FFF beams have a role in motion management as a way to reduce BOT.
73 Opportunities for Future Work and Development The current work provides opportunities for future development includ ing c linical implementation of the fiber optic temperature based respiration sensor for radiotherapy . To be used for 4D CT simulation the respiration signal must send a trigger to the CT scanner once per breathing cycle to acquire a CT slice. For treatme nt , gated radiation delivery requires the respiration system to hold the linear accelerator radiation beam and enable it automatically when the breathing cycle is within the predetermined gating window . The temperature based sensor can interface with data acquisition devices using either an analog output or digital port. The temperature sensor data could be transmitted to exist ing respiratory gating software or a new respiratory gating platform co uld be developed. Before the fiber optic temperature sensor is used cl inically , it is important to verify the relationship between the measured breathing cycle and actual tumor motion. When using an external signal as a respiratory surrogate, such as a temperature signal, the relationship with the internal target should be established. Future research on the fiber optic temperature sensor could include sampling the target position fluoroscopically for brief periods of time at a number of intervals to verify the relationship between the measured breathing cycle and actual tu mor motion. The relationship between the measured breathing cycle and actual tumor motion will determine how much dose delivery is improved by utilizing the temperature based respiration system .
74 LIST OF REFERENCES 1. Coia L, Schultheiss TE, Hanks GE (1 995) A Practical Guide to CT Simulation. Advanced Medical Publishing, Wisconsin 2. Timmerman R, Paulus R, Choy H (2010) SBRT for early stage lung cancer. J Amer Med Associat 303:1070 1076 3. Timmerman R, Hu C, Michalski J et al. (2014) Long term results of RTO G 0236: A phase II trial of SBRT in the treatment of patients with medically inoperable stage I non small cell lung cancer. ASTRO Annual Meeting, Abstract 56 4. Ross CS, Hussey DH, Pennington EC et al. (1990) Analysis of movement of intrathoracic neoplasms using ultrafast computerized tomography. Int J Radiat Oncol Biol Phys 18(3):671 677 5. Korin HW , Ehman RL, Riederer SJ et al. (1992) Respiratory kinematics of the upper abdomina l organs: a quantitative study. Magn Reson Med 23(1):172 178 6. Davies SC , Hill AL , Holmes RB et al. (1994) Ultrasound quantitation of respiratory org an motion in the upper abdomen. Br J Radiol 67(803):1096 1102 7. Kubo HD, Hill BC (1996) Respiration gated radiotherapy treatment: a technical study. Phys Med Biol 41(1):83 91 8. Balter JM, L am KL, McGinn CJ et al. (1998) Improvement of CT based treatment planning models of abdominal targets using static exhale imaging. Int J Radiat Oncol Biol Phys 41(4):939 943 9. Barnes EA , Murray BR , Robinson DM et al. (2001) Dosimetric evaluation of lung tu mor immobilization using b reath hold at deep inspiration. Int J Radiat Oncol Biol Phys 50(4):1091 1098 10. Stevens CW , Munden RF , Forster KM et al. (2001) Repiratory driven lung tumor motion is independent of tumor size, tumor lo cation, and pulmonary functio n. Int J Radiat Oncol Biol Phys 51(1):62 68 11. Kea l l PJ, Mageras GS , Balter JM et al. (2006) The management of respiratory motion in radiation oncology report of AA PM Radiation Task Group No. 76. Med Phys 33:3874 12. Wong JW, Sharpe MB, Jaffray DA et al. (1999 ) The use of active breathing control (ABC) to reduce margin for breathing motion. Int J Radiat Oncol Biol Phys 44(4):911 919
75 13. Kim DJ, Murray BR, Halperin R et al. (2001) Held breath self gating technique for radiotherapy of non small cell lung cancer: a feasibility study. Int J Radiat Oncol Biol Phys 49(1):43 49 14. Hanley J, Debois MM, Mah D et al. (1999) Deep inspiration breath hold technique for lung tumors: the potential value of target immobilization and reduced lung density in dose escalation. Int J Radiat Oncol Biol Phys 45(3):603 611 15. Pan T, Lee TY, Rietzel E et al. (2004) 4D CT imaging of a volume influenced by respiratory motion on multi slice CT. Med Phys 31(2):333 340 16. Mageras GS, Yorke E (2004) Deep inspiration breath hold and respiratory ga ting strategies for reducing organ motion in radiation treatment. Sem Radiat Oncol 14(1):65 75 17. Lagerwaard FJ, van Sornsen de Koste JR, Nijssen Visser MR et al. (2001) plannin g. Int J Radiat Oncol Biol Phys 51(4):932 937 18. van Sornsen de Koste JR, Lagerwaard FJ, Schuchhard Schipper RH et al. (2001) Dosimetric consequences of tumor mobility in radiotherapy of stage I non small cell lung cancer an analysis of data generated usin Radiother Oncol 61(1):93 99 19. de Koste JR, Lagerwaard FJ, de Boer HC et al. (2003) Are multiple CT scans required for planning curative radiotherapy in lung tumors of the lower lobe? Int J Radiat Oncol Biol Phys 55(5):1394 1399 20. Balter JM, Lam KL, McGinn CJ et al. (1998) Improvement of CT based treatment planning models of abdominal targets using static exhale imaging. Int J Radiat Oncol Biol Phys 41(4):939 943 21. Aruga T, Itami J, Aruga M et al. (2000) Target volume definition for upper abdominal irradiation using CT scans obtained during inhale and exhale phases. Int J Radiat Oncol Biol Phys 48(2):465 469 22. Mageras GS, Pevsner ED, Rosenzweig KE et al. (2004) Measurement of lung tumor motion using respiration correlated CT. Int J Radia t Oncol Biol Phys 60(3):933 941 23. Sonke J, Zijp L, Remeijer P et al. (2005) Respiratory correlated cone beam CT. Med Phys 32(4): 1176 1186
76 24. Lu W, Parikh PJ, El Naqa IM et al. (2005) Quantitation of the reconstruction quality of a four dimensional computer d tomography process for lung cancer patients. Med Phys 32(4): 890 901 25. Rietzel E, Chen GT, Choi NC et al. (2005) Four dimensional image based treatment planning: Target volume segmentation and dose calculation in the presence of respiratory motion. Int J Radiat Oncol Biol Phys 61(5): 1535 1550 26. Negoro Y , Nagata Y , Aoki T et al. (2001) The effectiveness of an immobilization device in conformal radiotherapy for lung tumor: reduction of respiratory tumor movement and evaluati on of the daily setup accuracy . Int J Radia t Oncol Biol Phys 50(4):889 898 27. Lohr F , Debus J, Frank C et al. (1999) Noninvasive patient fixation for extracra nial stereotactic radiotherapy. Int J Radia t Oncol Biol Phys 45(2):521 527 28. Blomgren H , Lax I, Naslund I et al. (1995) Stereotac tic high dose fraction radiation therapy of extracranial tumors using an accelerator. Clinical experience of the first thirty one patients. Acta Oncol 34(6):861 870 29. Thomas EM, P opple RA, Prendergast BM et al. (2013) Effects of flattening filter free and v olumetric modulated arc therapy delivery on treatment efficiency. J Appl Clin Med Phys 14(6):155 166 30. Zhao B, Yang Y, Li T et al. (2011) Statistical analysis of target motion in gated lung stereotactic body radiation therapy. Phys Med Biol 56:1385 1395 31. Kim S, Akpati HC, Kielbasa JE et al. (2004) Evaluation of intrafraction patient movement for CNS and head and neck IMRT. Med Phys 31(3):500 506 32. Cashmore J (2008) The characterization of unflattened photon beams from a 6 MV linear accelerator. Phys Med Bio l 53(7):1933 1946 33. Stath akis S, Esquivel C, Gutierrez A et al. (2009) Treatment planning and delivery of IMRT using 6 and 18MV photon beams without flat tening filter. Appl Radiat Isot 67(9):1629 1637 34. Georg D, Knoos T, McClean B (2011) Current status and f uture perspective of flattening filter free photon beams. Med Phys 38(3):1280 1293 35. Kragl G, Baier F, Lutz S et al. (2011) Flattening filter free beams in SBRT and IMRT: dosimetric assessment of peripheral doses. Z Med Phys 21(2):91 101 36. Kretschmer M, Saba tino M, Blechschmidt A et al. (2013) The impact of flattening filter free beam technology on 3D conformal RT. Radiat Oncol 8:133
77 37. Karzmark CJ, Nunan CS, Tanabe E (1993) Medical Electron Accelerators. McGraw Hill, Inc., Health Professions Division, New Yor k 38. Boucher L, Rodrigue S, Lecomte R et al. (2004) Respiratory gating for 3 Dimensional PET of the thorax: feasibility and initial results. J Nucl med 45(2):214 219 39. Yoo WJ, Jang KW, Seo JK et al . (2010) Development of respiration sensors using plastic o ptical fiber for respirator y monitoring inside MRI system. J Opt Soc Kor 14(3): 235 239 40. Yoo WJ, Jang KW, Seo JK et al . (2010) Development of nasal cavity and abdomen attached fi b er optic respiration sensors. J Kor Phys Soc 57( 6 ): 1550 1554 41. Yoo WJ, Jan g KW, Seo JK et al . (2011) Development of optical fiber based respiration sensor for noni nvasive respiratory monitoring Opt Rev 18( 1 ): 132 138 42. Yoo WJ, Seo JK, Jang KW et al . (2011) Fabrication and comparison of thermochromic material based fiber optic se nsors for monit oring the temperature of water. Opt Rev 18( 1 ):144 148 43. Yoo WJ, Jang KW, Seo JK et al . (2011) Development of a 2 channel embedded infrared fiber optic temperature sensor usin g silver halide optical fibers. Sens 11 : 9549 9559 44. Lu H, Brett R, Sharp G et al. (2007) The development and commissioning of a respiratory gated treatment system for proton therapy. IFMBE Proc 14:2215 2218 45. American College of Radiology, Magnetic Resonance Imaging Quality Control Manual (American College of Radiology, Reston, VA, 2004) 46. Ong CL , Verbakel WF, Cuiipers JP et al. (2010) Stereotactic radiotherapy for peripheral lung tumors: a comparison of volumetric modulated arc therapy with 3 other d elivery techniques. Radiother Oncol 97(3): 437 42 47. Zhang GG , Ku L, Dill ing TJ et al. (2011) Volumetric modulated arc planning for lung stereotactic body radiotherapy using conventional and unflattened photon beams: a dosimetric compari son with 3D technique. Radiat Oncol 6: p. 152 48. Jiang SB , Pope C, Al Jarrah KM et al. (2003) An experimental investigation on intra fractional organ motion effects in lung IMRT treatments. Phys Med Biol 48(12): p. 1773 84
78 49. Berbeco R I, Pope CJ , and Jiang SB (2006) Measurement of the interplay effect in lung IMRT treatment using EDR2 films. J Appl Clin Med 7(4): p. 33 42 50. Duan J , Shen S, Fiveash JB et al. (2006) Dosimetric and radiobiological impact of dose fractionation on respiratory motion induced IMRT delivery errors: a volumetric dose measurement study. Med Phys 33(5): p. 1380 7 51. Vrancic C , Tro fimov A, Chan TC et al. (2009) Experimental evaluation of a robust optimization method for IMRT of moving targets. Phys Med Biol 54(9): p. 2901 14 52. Court LE , Seco J, Lu XQ et al. (2010) Use of a realistic breathing lung phantom to evaluate dose delivery er rors. Med Phys 37(11): p. 5850 7 53. Ong C , Verbakel WF, Cuiipers JP et al. (2011) Dosimetric impact of interplay effect on RapidArc lung stereotactic treatment delivery. Int J Radiat Oncol Biol Phys 79(1): p. 305 11 54. Hrbacek J, Lang S, Graydon SN et al. (201 4) Dosimetric comparison of flattened and unflattened beams for stereotactic ablative radiotherapy of stage I non small cell lung cancer. Med Phys 41(3): p. 031709 55. Riley C , Yang Y, Li T et al. (2014) Dosimetric evaluation of the interp lay effect in respir atory gated RapidArc radiation therapy. Med Phys 41(1): p. 0 11715 56. Falk M, Pommer T, Keall P et al. (2014) Motion management during IMAT treatment of mobile lung tumors -a comparison of MLC tracking and gated delivery. Med Phys 41(10): p. 101707 57. Betzel G , Yi BY, Niu Y, Yu CX (2012) Is RapidArc more susceptible to delivery uncertainties than dynamic IMRT? Med Phys 39(10):5882 90 58. Li X , Yang Y, Li T et al. (2013) Dosimetric effect of respiratory motion on volumetric modulated arc therapy based lung SBRT treatment delivered by TrueBeam machine with flattening filter free beam. J Appl Clin Med Phys 14(6):4370 59. Timmerman R, Galvin J, Michalski J et al. (2006) Accreditation and quality assurance for Radiation Therapy Oncology Group: Multicenter clinical trial s using Stereotactic Body Radiation Therapy in lung cancer. Acta Oncol 45(7):779 786 60. Takeda A, Kunieda E, Sanuki N et al. (2009) Dose distribution analysis in stereotactic body radiotherapy using dynamic conformal multiple arc therapy. Int J Radiat Oncol Biol Phys 74(2):363 369
79 61. Ross CC, Kim JJ, Chen ZJ et al. (2011) A novel modified dynamic conformal arc technique for treatment of peripheral lung tumors using stereotactic body radiation therapy. Pract Radiat Oncol 1(2):126 134 62. Shi C, Tazi A, Fan g DX, Ian nuzzi C (2013) Implementation and evaluation of modified dynamic conformal arc (MDCA) technique for lung SBRT patients following RTOG protocols. Med Dosim 38(3):287 290 63. Radiation Therapy Oncology Group 0915 Protocol (2011) A Randomized Phase II Study Com paring 2 Stereotactic Body Radiation Therapy (SBRT) Schedules for Medically Inoperable Patients with Stage I Peripheral Non Small Cell Lung Cancer. Available at: http://rtog.org 64. Radiation Therapy Oncology Group 0813 Protocol (2013) Seamless Phase I/II St udy of Stereotactic Lung Radiotherapy (SBRT) for Early Stage, Centrally Located, Non Small Cell Lung Cancer (NSCLC) in Medically Inoperable Patients. Available at: http://rtog.org 65. Kragl G, af Wetterstedt S, Knausl B et al . (2009) Dosimetric characteristi cs of 6 and 10 MV unflattened photon beams. Radiother Oncol 93(1):141 146 66. Ponisch F, Titt U, Vassiliev ON et al . (2006) Properties of unflattened photon beams shaped by a multileaf collimator. Med Phys 33(6):1738 1746 67. Yarahma di M, Allahverdi M, Neda ie HA et al . (2013) Improvement of the penumbra for small radiosurgical fields using flattening filter free low megavoltage beams. Med Phys 23(4):291 299 68. Khoo VS, Oldham M, Adams EJ et al. (1999) Comparison of intensity modulated tomotherapy with stere otactically guided conformal radiotherapy for brain tumors. Int J Radiat Oncol Biol Phys 45(2):415 42 5 69. Murphy MJ, Chang S, Gibbs I et al. (2001) Image guided radiosurgery in the treatment of spinal metastases. Neurosurg Focus 11(6):e6 70. Josipovic M, Pers son GF, Logadottir A et al . (2012) Translational and rotational intra and inter fractional errors in patient and target position during a short course of frameless stereotactic body radiotherapy. Acta Oncol 51(5):610 617 71. ACR ASTRO practice guideline for the performance of stereotactic body radiation therapy. Practice Guidelines and Technical Standards. (American College of Radiology, Reston, VA, 2009)
80 BIOGRAPHICAL SKETCH Ashley Gale Smit h was born in Gainesville, FL to Steven and Kathy Gale . Sh e i s one of three children, along with sisters Shannon and Heather . Sh e attended high school in Frankfort , Kentucky at Franklin Count y High School. She then went to college at Indiana University, earning her and astrophysics in 200 0. She earned her maste Kentucky, and became employed as a medical physicist at Mayo Clinic in 2005. With an interest in supplementing her clinical career as a medical physicist with education and researc h , Ashley continued her studies at the University of Florida, where she earned her doctorate in biomedical engineering in 2015 . Ashley currently lives in Jacksonville, FL with husband Jason and dog Callie.