Image-guided adaptive radiation therapy : retrospective study and assessment of clinical workflow

Transcription

1 The University of Toledo The University of Toledo Digital Repository Theses and Dissertations 2013 Image-guided adaptive radiation therapy : retrospective study and assessment of clinical workflow Jason Michael Hudson The University of Toledo Follow this and additional works at: Recommended Citation Hudson, Jason Michael, "Image-guided adaptive radiation therapy : retrospective study and assessment of clinical workflow" (2013). Theses and Dissertations This Thesis is brought to you for free and open access by The University of Toledo Digital Repository. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of The University of Toledo Digital Repository. For more information, please see the repository's About page.

2 A Thesis entitled Image-Guided Adaptive Radiation Therapy: Retrospective Study and Assessment of Clinical Workflow by Jason Michael Hudson Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master of Science in Biomedical Science Degree in Medical Physics David Pearson Ph.D., Committee Chair E. Ishmael Parsai Ph.D., Committee Member Diana Shvydka Ph.D., Committee Member Patricia R. Komuniecki Ph.D., Dean College of Graduate Studies The University of Toledo August 2013

3 Copyright 2013, Jason Michael Hudson This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author ii

4 An Abstract of Image-Guided Adaptive Radiation Therapy: Retrospective Study and Assessment of Clinical Workflow by Jason Hudson Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master of Science in Biomedical Science Degree in Medical Physics The University of Toledo August 2013 Patients undergoing intensity modulated radiation therapy are simulated prior to treatment using computed tomography. This simulation is used to develop a single treatment plan that will be used each day to deliver the prescribed dose of radiation. Daily volume changes of critical structures which may deviate from the volumes observed at the time of simulation, specifically in the case of prostate cancer patients, may alter the delivered dose to the planning target volume. The delivered dose to the prostate, seminal vesicles, rectum, bladder, and femoral heads have been re-calculated on daily cone-beam computed tomography setup images acquired prior to each treatment. The deviation from the planned dose to these critical structures has been evaluated on the basis of dose-volume statistics commonly used at this institution in the case of two prostate cancer patients, both on a daily basis and over the entire course of treatment in the way of accumulated dose. The average daily difference in the dose to ninety-five percent of the prostate from the planned dose was observed to be -1.40% and -0.42% for the primary and boost doses of radiation, respectively, in a prostate patient with standard iii

5 margins. The average daily difference in the dose to ninety five percent of the prostate was observed to be -1.13% over the course of treatment for a patient prescribed one dose of radiation with non-standard margins. iv

6 Acknowledgements I would like to express my sincere gratitude to my research advisor, Dr. Pearson for his input and guidance, as well as my committee members Dr. Parsai and Dr. Shvydka for their support and patience during the completion of this research. I would like to thank the faculty and staff here in the department of Radiation Oncology at UTMC, as well as my classmates for making my time here enjoyable. v

7 Table of Contents Abstract... iii Acknowledgements... v Table of Contents... vi List of Tables... ix List of Figures... xi List of Abbreviations... xiv List of Symbols... xv 1. Introduction Prostate IMRT Prostate Localization Techniques Prostate Localization, 4D Prostate Localization, 3D Prostate Localization, 2D Imaging in Radiation Therapy Computed Tomography Cone Beam Computed Tomography Planar Imaging vi

8 3. Adaptive Radiation Therapy Rationale for Adaptive Radiation Therapy Deformable Registration Methods and Materials Registration of Daily CBCT s Dose Recalculation and DVH Analysis Dose Accumulation and DVH Analysis Results Daily DVH vs. Planned DVH Prostate 1, Primary Prostate 1, Boost Prostate 1, Average Daily Deviation Prostate Prostate 2, Average Daily Deviation Accumulated DVH vs. Planned DVH Prostate 1, Primary Prostate 1, Boost Prostate 1, Composite Prostate Summary and Conclusions vii

9 6.1 Clinical Workflow Contour Subjectivity PTV Margins Future Directions References Appendix A Appendix B viii

10 List of Tables Table 3-1: CT# to Density Calibration Values Table 5-1: Daily Dose-Volume Parameters Table 5-2: Prostate 1, Daily Average Percent Difference from Plan, Primary Table 5-3: Prostate 2, Daily Average Percent Difference from Plan Table 5-4: Course Composite Dose Constraints Table 5-5: "Prostate 1", Accumulated Dose, Bladder, Primary Comparison Table 5-6: "Prostate 1", Accumulated Dose, Rectum, Primary Comparison Table 5-7: "Prostate 1", Accumulated Dose, LT Fem Head, Primary Comparison Table 5-8: "Prostate 1", Accumulated Dose, RT Fem Head, Primary Comparison Table 5-9: "Prostate 1", Accumulated Dose, Prostate, Primary Comparison Table 5-10: "Prostate 1", Accumulated Dose, Sem. Ves., Primary Comparison Table 5-11: "Prostate 1", Accumulated Dose, Bladder, Boost Comparison Table 5-12: "Prostate 1", Accumulated Dose, Rectum, Boost Comparison Table 5-13: "Prostate 1", Accumulated Dose, LT Fem Head, Boost Comparison Table 5-14: "Prostate 1", Accumulated Dose, RT Fem Head, Boost Comparison Table 5-15: "Prostate 1", Accumulated Dose, Prostate, Boost Comparison Table 5-16: "Prostate 1", Accumulated Dose, Seminal Vesicles, Boost Comparison ix

11 Table 5-17: "Prostate 1", Accumulated Dose, Bladder, Composite Comparison Table 5-18: "Prostate 1", Accumulated Dose, Rectum, Composite Comparison Table 5-19: "Prostate 1", Accum. Dose, LT Fem Head, Composite Comparison Table 5-20: "Prostate 1", Accum. Dose, RT Fem Head, Composite Comparison Table 5-21: "Prostate 1", Accumulated Dose, Prostate, Composite Comparison Table 5-22: "Prostate 2", Accumulated Dose, Bladder Table 5-23: "Prostate 2", Accumulated Dose, Rectum Table 5-24: "Prostate 2", Accumulated Dose, LT Fem Head Table 5-25: "Prostate 2", Accumulated Dose, RT Fem Head Table 5-26: "Prostate 2", Accumulated Dose, Prostate Table A-1: "Prostate 1", Daily % Difference from Planned, Primary Table A-2: "Prostate 1", Daily % Difference from Planned, Primary Table A-3: "Prostate 1", Daily % Difference from Planned, Boost Table A-4: "Prostate 1", Daily % Difference from Planned, Boost Table A-5: "Prostate 2", Daily % Difference from Planned Table A-6: "Prostate 2", Daily % Difference from Planned x

12 List of Figures Fig 1-1: Source Coils and Receiver Coils... 4 Fig 1-2: AP and LT Lat View of Localization Markers... 6 Fig 1-3: BAT Ultrasound System... 6 Fig 1-4: BAT Ultrasound Probe and Arm... 7 Fig 2-1: CT Scanner and Patient Geometry... 9 Fig 2-2: Geometry of MV/kV Sources and Detector Panels Fig 2-3: CBCT kv Source and Detector Panel Geometry Fig 3-1: C T# to Density Calibration, Varian OBI vs. Philips TPS Fig 4-1: Selection of CBCT's and Adaptive Recontour Deform Workflow Fig 4-2: "Contour CoPilot" Tool Fig 4-3: "Contour CoPilot" Editing Process Fig 4-4: Selection of the "Dose-Accumulation - Deformable" Workflow Fig 4-5: "Dose Accumulation - Deformable" Dose Warning Fig 5-1: "Prostate 1", Daily DVH vs. Planned DVH, Bladder, Primary Fig 5-2: "Prostate 1", Daily DVH vs. Planned DVH, Rectum, Primary Fig 5-3: "Prostate 1", Daily DVH vs. Planned DVH, LT Femoral Head, Primary Fig 5-4: "Prostate 1", Daily DVH vs. Planned DVH, RT Femoral Head, Primary xi

13 Fig 5-5: "Prostate 1", Daily DVH vs. Planned DVH, Prostate, Primary Fig 5-6: "Prostate 1", Daily DVH vs. Planned DVH, Seminal Vesicles, Primary Fig 5-7: "Prostate 1", Daily DVH vs. Planned DVH, Bladder, Boost Fig 5-8: "Prostate 1", Daily DVH vs. Planned DVH, Rectum, Boost Fig 5-9: "Prostate 1", Daily DVH vs. Planned DVH, LT Femoral Head, Boost Fig 5-10: "Prostate 1", Daily DVH vs. Planned DVH, RT Femoral Head, Boost Fig 5-11: "Prostate 1", Daily DVH vs. Planned DVH, Prostate, Boost Fig 5-12: "Prostate 1", Daily DVH vs. Planned DVH, Seminal Vesicles, Boost Fig 5-13: "Prostate 2", Daily DVH vs. Planned DVH, Bladder Fig 5-14: "Prostate 2", Daily DVH vs. Planned DVH, Rectum Fig 5-15: "Prostate 2", Daily DVH vs. Planned DVH, LT Femoral Head Fig 5-16: "Prostate 2", Daily DVH vs. Planned DVH, RT Femoral Head Fig 5-17: "Prostate 2", Daily DVH vs. Planned DVH, Prostate Fig 5-18: "Prostate 1", Accum. DVH vs. Planned DVH, Bladder, Primary Fig 5-19: "Prostate 1", Accum. DVH vs. Planned DVH, Rectum, Primary Fig 5-20: "Prostate 1", Accum. DVH vs. Planned DVH, LT Femoral Head, Primary Fig 5-21: "Prostate 1", Accum. DVH vs. Planned DVH, RT Femoral Head, Primary Fig 5-22: "Prostate 1", Accum. DVH vs. Planned DVH, Prostate, Primary Fig 5-23: "Prostate 1", Accum. DVH vs. Planned DVH, Seminal Vesicles, Primary Fig 5-24: "Prostate 1", Accum. DVH vs. Planned DVH, Primary Fig 5-25: "Prostate 1", Accum. DVH vs. Planned DVH, Bladder, Boost Fig 5-26: "Prostate 1", Accum. DVH vs. Planned DVH, Rectum, Boost Fig 5-27: "Prostate 1", Accum. DVH vs. Planned DVH, LT Femoral Head, Boost xii

14 Fig 5-28: "Prostate 1", Accum. DVH vs. Planned DVH, RT Femoral Head, Boost Fig 5-29: "Prostate 1", Accum. DVH vs. Planned DVH, Prostate, Boost Fig 5-30: "Prostate 1", Accum. DVH vs. Planned DVH, Seminal Vesicles, Boost Fig 5-31: "Prostate 1", Accumulated DVH vs. Planned DVH, Boost Fig 5-32: "Prostate 1", Accum. DVH vs. Planned DVH, Bladder, Composite Fig 5-33: "Prostate 1", Accum. DVH vs. Planned DVH, Rectum, Composite Fig 5-34: "Prostate 1", Accum. DVH vs. Planned DVH, LT Fem Head, Composite Fig 5-35: "Prostate 1", Accum. DVH vs. Planned DVH, RT Fem Head, Composite Fig 5-36: "Prostate 1", Accum. DVH vs. Planned DVH, Prostate, Composite Fig 5-37: "Prostate 1", Accum. DVH vs. Planned DVH, Seminal Vesicles, Composite. 61 Fig 5-38: "Prostate 2", Accum. DVH vs. Plan DVH, Bladder Fig 5-39: "Prostate 2", Accum. DVH vs. Planned DVH, Rectum Fig 5-40: "Prostate 2", Accum. DVH vs. Planned DVH, LT Fem Head Fig 5-41: "Prostate 2", Accum. DVH vs. Planned DVH, RT Fem Head Fig 5-42: "Prostate 2", Accum. DVH vs. Planned DVH, Prostate Fig 6-1: Adaptive Therapy Workflow xiii

15 List of Abbreviations ART...Adaptive Radiation Therapy CBCT...Cone Beam Computed Tomography CT...Computed Tomography CTV...Clinical Target Volume DVH...Dose-Volume Histogram DRR...Digitally Reconstructed Radiograph EBRT...External Beam Radiation Therapy EPID...Electronic Portal Imaging Device GTV...Gross Tumor Volume IGRT...Image-Guided Radiation Therapy IMRT...Intensity-Modulated Radiation Therapy kvp...kilovoltage Peak mas...milliampere-second PTV...Planning Target Volume xiv

16 List of Symbols cgy...centigray D xx...dose to (xx)% Fx...Fraction xv

17 Chapter 1: Introduction Among American men, prostate cancer is the second-most prevalent type of cancer. (1) In 2013 alone, about 238,590 new diagnoses of prostate cancer are predicted, with about 29,720 of those diagnoses resulting in the death of the diagnosed. In every six men, one will be diagnosed with prostate cancer during the course of their lifetime. As a result of these startling statistics, much has been done in the way of medical research and development of various treatment options for victims of prostate cancer. One of the most common modalities of treatment is external beam radiation therapy (EBRT). More specifically, the form of EBRT most commonly used is intensity-modulated radiation therapy (IMRT), which will be discussed later in this chapter. Daily volume changes that occur in the bladder and rectum cause the prostate to be moved from its original position at the time of simulation. The volume changes that come as a result of differences in bladder and rectal filling cause the dose delivered to these avoidance structures to change, and an effort to quantify these changes in dose delivered to the bladder and rectum as well to the femoral heads and clinical target volume (CTV) are included in this research. Two fractionation schemes and expansions from the CTV will be examined in two different prostate cancer patients. 1

18 The goal of this research is to quantify the dose difference observed in the daily treatment and entire course of treatment in two prostate cancer patients. Also, a goal of this research is to examine the feasibility of dose recalculation and re-optimization in the adaptive treatment of these patients. 1.1 Prostate IMRT Patients undergoing intensity-modulated radiation therapy to the prostate typically receive a dose of 4500 cgy to 5040 cgy over the course of 5 weeks for primary treatment to the planning target volume (PTV), which contains the clinical target volume (CTV), drawn by the physician on a treatment-planning computed tomography (CT) image, plus a margin. Once the primary course of radiation is completed, the patient then receives a boost dose of radiation, typically bringing the dose to the prostate up to around 7700 cgy. In most cases, the CTV is composed of the prostate and seminal vesicles. In cases where the patient is not prescribed a boost dose, the patient receives approximately 7000 cgy to the PTV, which is comprised of the CTV plus a margin. In cases such as these, the CTV is composed only of the prostate. At our institution, the PTV to which a primary dose of radiation is prescribed is typically expanded from the CTV by 1 centimeter in the superior-inferior, left-right lateral, and anterior directions, while an expansion of 7millimeters is used to form the posterior aspect of the PTV. The PTV to which a boost dose of radiation is prescribed is comprised of the prostate plus a 7 millimeter margin in the superior-inferior, anterior, and left-right lateral directions, while an expansion of 5 millimeters is used to form the posterior aspect of the boost PTV. For cases in which the patient receives no boost dose of radiation, but instead receives the entire prescription to only the prostate, the margin 2

19 used to define the PTV is a non-standard one, with an expansion of 5 millimeters in all directions, with no margin at the posterior aspect of the PTV. The purpose of a PTV in EBRT is to ensure that the CTV or gross tumor volume (GTV) around which the PTV was formed receives full dose during each fraction of treatment (7). The PTV ensures that even in instances of minute differences in the translational or rotational setup of the patient each day, the CTV will receive the full dose of the prescription as long as the treatment plan is optimized to deliver the prescribed dose to an approved percentage of the PTV (typically 95%). The PTV also ensures that the CTV gets full dose if there are structures close to the prostate, such as the bladder and rectum, which frequently change in volume and may cause the CTV to shift in the pelvis. These shifts in the body are compensated for through the use of in-room imaging including both planar and volumetric forms of imaging. 1.2 Prostate Localization Techniques As with all forms of radiation therapy, the accurate localization of the prostate on a day-to-day basis is very important. Without proper localization, the planned dose may differ from what is actually delivered to the treatment area. On any given day, without proper localization, the dose delivered to avoidance structures around the prostate may be increased due to inaccurate localization. Several forms of localization exist, including 4D, 3D, and 2D localization techniques Prostate Localization, 4D In a 4D localization technique, intrafractional movement is observed. Intrafractional movement comes as a result of the influence of nearby avoidance 3

20 structures including the bladder and rectum which may change volume during a delivered fraction of radiation. The most common method for 4D prostate localization is through the use of radiofrequency transponders. (6) These transponders are surgically implanted into the prostate of the patient, and the radiofrequency of these small, glass encased transponders is detected during treatment. A magnetic source and array of receiver coils are used to track the position of the implanted transponders. This array of receiver coils is tracked by in-room infrared tracking, forming a secondary frame of reference to the frame of reference formed by the position of the transponders in relation to the receiver coil array. The combination of these two continually updated (rate of 10 Hz) frames of reference result in the continuous feedback of the location of the prostate for the use of accurate patient positioning and position information of the prostate during the delivered treatment. A schematic of the source coils, and receiver coil array adapted from Balter et. al. (6) is shown here: Fig 1-1: Source Coils and Receiver Coils 4

21 1.2.2 Prostate Localization, 3D 3D prostate localization techniques are the techniques most commonly used in a prostate IMRT treatment regimen. Integral in the 3D prostate process is the use of inroom imaging. This type of imaging is called cone-beam computed tomography (CBCT) and will be described in greater detail in the next chapter of this document. The use of CBCT images allows the clinician to assess the position of the prostate in the patient prior to treatment. At our institution, a Varian (Varian Medical Systems, Palo Alto) onboard imager (OBI) is used for this purpose. Patients are aligned to the tattoos made at the time of simulation, and a CBCT image is acquired. After the image is acquired, a 3D/3D match is made using the OBI workstation. The shifts from the acquisition isocenter (the isocenter of the CBCT) are made to align the prostate to the prostate on the treatment planning CT. As an aide in the process of soft-tissue matching, some institutions employ the use of implanted gold markers for localization. These gold markers are arranged in such a way that the position of the prostate can be verified on axial, sagittal, and coronal slices of the acquired CBCT scan. Below is an example of the seeds as viewed by anterior-posterior (AP) and left lateral (LT Lat) setup fields, with the seeds as they are viewed on a digitally reconstructed radiograph (DRR) circled in red: 5

22 AP LT Lat Fig 1-2: AP and LT Lat View of Localization Markers A separate method for the 3D localization of the prostate that does not involve the use of CBCT is through the use of ultrasound. The BAT (B-mode Acquisition and Targeting) system depicted below (NOMOS Corporation, Sewickley, PA) was formerly used for prostate localization at our institution Fig 1-3: BAT Ultrasound System 6

23 The BAT system makes use of an ultrasound probe for pre-treatment prostate localization. This probe is mounted to an arm which is given a frame of reference defined by defined by the current table position as it relates to the position of the arm of the BAT system, and by the current gantry position as it relates to the position of the arm of the BAT system. Fig 1-4: BAT Ultrasound Probe and Arm With this coordinate system defined, the BAT system is able to translate the approved position of the prostate detected by the ultrasound into table shifts that can be performed to re-center the prostate to the isocenter of the planned treatment fields Prostate Localization, 2D In a 2D prostate localization technique, patients are implanted with gold markers for localization in much the same way as if they were to be localized by a 3D localization technique. The implanted seeds are used to align the prostate to orthogonal digitally reconstructed radiographs (DRRs) generated from the treatment planning CT. The 7

24 acquired orthogonal images either by kv or MV imaging techniques utilize an image acquired by the gantry-mounted electronic portal imaging device (EPID) or kv imaging panel, and oppositely adjacent kv/mv source to perform an on-line match similar to the match made in 3D localization techniques. After an acceptable online match is made, the shift of the acquired image to the planned image is produced in the online matching software at the treatment console of the linear accelerator. These shifts are translated into table shifts and are sent to the treatment console in order to position the patient as closely as possible to the isocenter of the planned treatment fields. 8

25 Chapter 2: Imaging in Radiation Therapy 2.1 Computed Tomography In radiation therapy, several forms of imaging are used to determine the extent of disease in a cancer patient and to localize the areas to be treated. The most important types of images in radiation therapy are those generated by means of computed tomography (CT). CT images are produced by an x-ray generating source and oppositely adjacent ring of detectors. This x-ray generating source and ring of detectors rotate about a patient as they are passed through the central bore, along the z-axis of the CT scanner. This type of acquisition is called a helical scan. Images are continually taken as the patient is moved through the scanner, thereby forming a helix along the patient s z-axis. Gantry X-rays Patient is moved along z-axis of scanner Scanner Bore Patient is raised to center of CT scanner field of view Fig 2-1: CT Scanner and Patient Geometry 9

26 Modern CT scanners are considered to be multi-detector computed tomography (MDCT) scanners. The detector panel oppositely adjacent to the x-ray tube in an MDCT is composed of an array of detectors, giving rise to narrow cone beam geometry. (2) In our clinic, a 16 slice Philips Big Bore TF PET/CT (Philips, Koninklijke N.V.) is used for scans of cancer patients. A 16-slice scanner has 16 detector arrays located in an arc oppositely adjacent to the kv source. MDCT scanners use reconstruction algorithms to convert the signal given by the 80 to 120 kv spectra (effective energy of 40 to 60 kev) (2) into an image of the patient anatomy after these photons have passed through the patient. These CT images are the primary image format used in the development of a treatment plan in radiation oncology. The CT images are exported to a treatment planning system where the clinician can draw target volumes and avoidance structures. Traditionally, it is this treatment plan that a patient will be treated with over the entire course of treatment, with the use of in-room imaging to verify patient setup and target location in an image-guided radiation therapy (IGRT) paradigm. 2.2 Cone Beam Computed Tomography Cone beam computed tomography (CBCT) is one of the most common forms of in-room computed tomography. Most modern linear accelerators are capable of this type of imaging for the purpose of IGRT. CBCT utilizes a kv source of radiation and a detector panel that are orthogonal to the treatment MV source and detector panel, as illustrated in Fig. 2-2: 10

27 Gantry Imaging kv x- rays MV Source Therapeutic MV X-rays kv Source (Patient) kv Detector Panel Isocenter MV Detector Panel Fig 2-2: Geometry of MV/kV Sources and Detector Panels In a CBCT acquisition, the patient is centered to the treatment beam isocenter, and the gantry makes one rotation around the patient while the kv source emits radiation with kvp and mas specific to the imaging protocol selected for the treatment site (pelvis, head, etc.). Opposite to the kv and MV source are amorphous silicon detector panels. Both detector panels are normal to the path of the incident radiation. The detector panels and kv source are capable of being retracted during treatment, as they are mounted on retractable arms. The images acquired before treatment allow the radiation therapy team to compare the treatment planning CT with the cone beam CT and make necessary shifts to adequately cover the target volume to be treated. This basic process defines the Image 11

28 Guided Radiation Therapy (IGRT) clinical workflow, where an in-room image is used to align the patient properly, using external markings made at the time of the simulation to acquire an image. Therapists are instructed to make shifts necessary to align the patient based on anatomy, take additional images to verify the shifts, and treat the patient upon approval of the final set of images. Unlike the standard helical MDCT scanner, CBCT makes use of full cone beam geometry. (2) In full cone beam geometry, the fan angle that is used by the imaging system is almost the same as the cone angle. The source-detector geometry is depicted below: Fig 2-3: CBCT kv Source and Detector Panel Geometry 2.3 Planar Imaging Modern accelerators not only use the imaging detector panels for CBCT, but also use detector panels for the acquisition of planar images. At our institution, patients not undergoing treatment for prostate cancer or breast cancer receive both CBCT as well as planar images. Planar images are taken instead of CBCT images on certain days in order 12

29 to reduce the imaging dose to the patient. Two types of planar images may be acquired, including megavoltage (MV) or kilovoltage (kv) images. MV images allow the radiation therapy team to review a portal image of the treatment fields. In a portal image, the shaped treatment field is used to expose the patient, often times followed by a larger open field to verify that the site to be treated is within the planned treatment fields. MV planar images for verification are particularly useful when a thoracic site is to be treated and the location of the trachea and trachea carina must be verified. Kilovoltage images offer superior resolution when verifying bony anatomy due to an increased photoelectric effect upon interaction with highly attenuating bone. 13

30 Chapter 3: Adaptive Radiation Therapy 3.1 Rationale for Adaptive Radiation Therapy Throughout the course of treatment it is likely that patient anatomy will change. Traditionally, patients are treated with a plan generated from one treatment planning CT, and are treated based on shifts made from in-room imaging. While this shift is made based on anatomy, the beam itself has not. Rather, the beam maintains its original qualities from the time of treatment planning. In an IMRT treatment, the fields themselves are still optimized for the target volume delineated on the treatment planning day as well as those volumes which were drawn as avoidance structures. In prostate IMRT, the volumes of the bladder and rectum change daily. As a result of these volume changes, the previously approved dose-volume histogram (DVH) is no longer a truly representative DVH of what dose is being delivered to the target volume and the nearby organs at risk (OAR) on a daily basis. In order to maintain the treatment planning DVH characteristics as closely as possible, Adaptive Radiation Therapy (ART) must be the new clinical protocol to maintain an acceptable dose level to both Organs at Risk (OAR s) as well as target volumes. ART requires periodic re-planning in order to maintain these dose levels. 14

31 CT # Periodic or daily re-imaging of the patient is required in order to generate a new treatment plan which considers the daily volume of the patient. In order to do so, the images used must be volumetric images of sufficient quality to yield accurate calibration curves for the calculation of dose. Using an anatomical site-specific CT number to density calibration curve for CBCT images may be necessary to achieve this accuracy. Studies show that when using a site-specific CT number to density calibration for CBCT images, a 2% dose agreement is observed between treatment planning CT and CBCT images in phantom studies. (3) The anatomical site-specific calibration performed in this research was for the Pelvis CBCT protocol of the Varian OBI using the Catphan 404 (The Phantom Laboratory, Salem, NY, USA). The results of that CT number to density calibration in comparison to the CT number to density calibration curve used in the treatment planning of all patients at our institution are given here: 9000 C T# to Density Calibration, Varian OBI vs. Philips TPS Varian OBI Philips TPS Density (g/cm3) Fig 3-1: C T# to Density Calibration, Varian OBI vs. Philips TPS 15

32 Density (g/cm 3 ) Varian OBI CT# Philips TPS CT# Absolute % Difference % % % % % % % % % Table 3-1: CT# to Density Calibration Values The results of this calibration indicate that for tissues with density between g/cm 3 and g/cm 3, the densities observed are in good agreement between the Varian OBI and the CT to density table used for treatment planning purposes. Given the convenience of CBCT to verify patient setup, the adaptive re-planning of a patient on CBCT images becomes a particularly appealing idea, especially when considering the similarity of the CT number to density calibration of CBCT images as compared to the CT number to density calibration of the treatment planning CT images. Through re-planning, OAR volume changes as well as volume change of the area being treated can be considered in the optimization of the fields used in IMRT. While quality of the treatment given to the patient is of the utmost importance, clinical throughput must not be sacrificed. In an effort to maintain the throughput of a clinic, the speed with which adaptive re-planning is done must be adequate so as to not extend the length of treatment for any one patient beyond which is reasonable (approx. 15 minutes). As an aide in the adaptive re-planning process, a quick deformation of the treatment 16

33 planning dose distribution to the newly acquired CBCT image may be helpful to determine if re-planning is necessary for any one fraction. 3.2 Deformable Registration In an ART workflow, deformable image registration techniques are particularly useful. Deformable image registration allows the user to accurately match the anatomy between two sets of volumetric images in the event that the patient experiences weight loss or if the target to be treated changes size/shape. Beyond an ART workflow, deformable image registration is particularly useful in the event that a patient has multiple images which must be aligned for treatment that may have been acquired at different times on different CT scanners. For instance, if the patient has a PET/CT performed on a CT scanner with a non-rigid couch surface, the acquired images can be deformed to a treatment planning CT that may have been performed on a CT scanner that has a rigid couch. At our institution, MIM Software s VoxAlign Deformation Engine (4) (MIM Software, Cleveland, OH), a commercially available deformable image registration algorithm is used. The VoxAlign deformation algorithm is a constrained, intensity-based, free-form deformable registration which uses the entire window and level of the CT contrast for matching the primary and secondary image sets. This is in sharp contrast to an approach known as a mutual information approach, which compresses the contrast in the CT, leaving less information for accurate tissue matching. The VoxAlign algorithm accurately captures local as well as global changes through the use of degrees of freedom which number in the millions. As an attempt to reduce the number of deformations that may be unexpected or unintended, the VoxAlign Algorithm is constrained so as to prevent producing deformations which deform bone or tear tissue. 17

34 In a study of the accuracy of the algorithm by J.W. Piper (5), correlation coefficients of rigid versus deformable registration were calculated for three patients exhibiting significant weight loss over the course of IMRT treatment. The investigator reports correlation coefficients of 0.890, 0.921, and after a rigid registration of pretreatment CT s with images acquired over the course of treatment. The same images pairs were deformably registered to each other for the same three patients. The investigator reports correlation coefficients of 0.979, 0.989, and 0.978, exhibiting a significant improvement in the correlation of the aligned secondary images of each patient to the patients primary image after the use of the VoxAlign deformable registration algorithm. These results indicate the efficacy of using this commercially available algorithm in cases of changes in anatomy that occur over the time elapsed between treatment planning CT acquisition and images acquired over the course of the treatment of the patient. 18

35 Chapter 4: Methods and Materials 4.1 Registration of Daily CBCT s Daily CBCT s of two prostate patients undergoing IMRT were exported from the Offline Review module in Varian ARIA record and verify to a local folder. From the local folder, these images were imported into a database in MIM Version Once loaded into the database, the CBCT from the first day of treatment was contoured. This CBCT was used to deform contours to the CBCT acquired on the second day of treatment using the Adaptive Recontour Deform workflow: Fig 4-1: Selection of CBCT's and Adaptive Recontour Deform Workflow 19

36 Upon initiating the Adaptive Recontour - Deform workflow, the first step is a rigid fusion of the daily CBCT s. The workflow pauses for review of the rigid fusion, where the user may adjust the alignment of the secondary image to the primary image before proceeding with the Adaptive Recontour - Deform workflow. After review, the rigid fusion was adjusted using the Box-Based Assisted Alignment when appropriate. The Box-Based Assisted Alignment tool allows the user to draw a box defined in axial, sagittal, and coronal views around rigid anatomy that is in close proximity to the treatment area (in this case, the prostate and surrounding rigid anatomy). Once a satisfactory region has been drawn, the user then clicks the green flag to proceed with the alignment, and the software does its best to match the rigid anatomy in the user-defined region. The Adaptive Recontour - Deform workflow is paused until an acceptable rigid fusion is made before proceeding with the next steps in the workflow. Once an acceptable rigid fusion is made, the Adaptive Recontour - Deform workflow proceeds with deforming the secondary image to the primary image set. To perform deformations, the software uses its proprietary VoxAlign algorithm described previously. After the workflow is completed and the deformation has been applied, the resultant contours produced by the deformation are reviewed and adjusted as necessary. One valuable tool used in the adjustment of the contours produced by the deformation is MIM s Contour CoPilot, depicted below: 20

37 Fig 4-2: "Contour CoPilot" Tool The Contour CoPilot is advantageous in that it provides the capability to perform an automatic, local image deformation. The Contour CoPilot assists in making contours based on the image deformation of the previously contoured slice. Unlike the Adaptive Recontour - Deform workflow, the Contour CoPilot is not constrained, meaning that the deformation is not limited by any threshold or gradient between types of tissues. This means that the deformation by which it produces contours is not set to avoid the deformation of bone, and does not attempt to produce a smooth deformation of skin or other tissue, as the segment of the image that is deformed is not used for alignment or localization purposes. Rather, the deformation produced by the Contour CoPilot is only used to make contours on immediately adjacent slices of the image which is being contoured. When a contour is generated by the Contour CoPilot, it first appears as a colorwash on the slice immediately adjacent to the previously contoured slice. If the User clicks on the slice, the colorwash contour turns into an editable contour: 21

38 Fig 4-3: "Contour CoPilot" Editing Process Should the user decide, the resultant contour created by clicking in the colorwash region may be edited. If the user does decide to edit the contour, the previous slices adjacent to this slice will show a suggested Contour CoPilot contour. Until the Contour CoPilot is turned off, it will continue to make suggestions on slices adjacent to the most recently contoured slice. 4.2 Dose Recalculation and DVH Analysis After all treatment day images (total of 71 CBCT s) were adaptively contoured and adjusted accordingly, these images were imported into the Pinnacle 3 (Philips, Koninklijke, N.V.) treatment planning system. In addition to the import of the daily CBCT images, the daily contours were also imported from MIM. The planned treatment fields were imported into the planning system via a script produced by the MU validation software RadCalc (LifeLine Software, Austin, TX, USA). A new prescription was added and set to deliver the same number of monitor units of the original treatment plan. The treatment fields, set to this prescription were edited to match the same beam weighting of 22

39 the original treatment fields. After computing the dose of these fields, a popular opensource script (8) was used to export the tabular DVH to a text file, which was later analyzed in Microsoft Excel (Microsoft Corporation, Redmond, WA, USA). A sample of the output of this script is included in Appendix B. The tabular DVH data imported into Excel were organized by region of interest and each region of interest was normalized such that the daily DVH data could be plotted alongside the planned DVH to observe the differences in the daily DVH s as opposed to the single fraction planned DVH. The same daily DVH data was used to produce a nearest-neighbor interpolation through the use of the daily data and an indexing array such that a standard percentage of the volume correlated with the absolute dose that corresponded to that portion of the volume. A resolution of 0.1% normalized volume was chosen, and the daily nearest neighbor interpolation values were compared with the planned single fraction DVH for specified in-house dose-volume constraints to be specified later. 4.3 Dose Accumulation and DVH Analysis The daily doses were exported back to MIM in order to utilize the Dose Accumulation - Deformable workflow (MIM Software, Cleveland, OH): 23

40 Fig 4-4: Selection of the "Dose-Accumulation - Deformable" Workflow Using MIM s deformable registration algorithm, the CBCT to CBCT deformation is applied to the daily doses to accumulate the daily doses on a voxel-tovoxel basis according to the current day s critical structures. Previous day CBCT s are fused to the current day CBCT scans. A rigid fusion is first reviewed to assure that the treatment area is correctly matched to the previous day CBCT. This rigid fusion is the same performed during the Adaptive Recontour Deform workflow used to create daily contours of the patient. During the Dose Accumulation - Deformable workflow, the workflow pauses to ask the user how many fractions each imported dose grid represents. This fraction is used to scale the imported dose grid according to the prescription dose that the dose grid represents. In the case of Daily CBCT scans, a 1 fraction dose of the patients respective prescription was used in a 1/1 fraction to max dose ratio: 24

41 Fig 4-5: "Dose Accumulation - Deformable" Dose Warning The distribution is initially set by the Dose Accumulation - Deformable workflow to be proportionate to the Max Dose detected by the software. The user may then set the isodose lines to be proportionate to the daily prescribed dose. After the Dose Accumulation - Deformable workflow was completed, a DVH was generated in MIM. The DVH generated corresponds to the current day s critical structures. By using the current critical structures, the effects of volume change as well as organ positioning with respect to the rigid anatomy are preserved. Through using the daily contours to assess the current DVH of the patient, the same surface areas of the bladder and rectum as well as the prostate, seminal vesicles (if included in the CTV of the patient), and femoral heads are plotted in the daily DVH, and a more accurate accumulated dose volume histogram can be evaluated. 25

42 In the case of a patient with a boost prescription, the primary and boost prescriptions were accumulated separately and then together to produce a course composite DVH of accumulated dose, using the Accumulate Dose, Deformable workflow. The course composite accumulated dose evaluated based on dose-volume constraints for bladder, rectum, and femoral heads used at our institution. In addition to these dose constraints, 95% of the CTV was evaluated for the dose it received each day in addition to the entire dose it received throughout the prescribed treatment. 26

43 Chapter 5: Results 5.1 Daily DVH vs. Planned DVH The treatment plans for two prostate patients, Prostate 1 and Prostate 2 were reproduced on daily CBCT images acquired before each treatment. The daily doses were observed and included on a DVH depicting the daily, one-fraction DVH on a perstructure basis. The deviation from the one-fraction planned DVH was calculated for dose-volume parameters in the form of a percent difference. The parameters evaluated were as follows, where D xx is the dose delivered to (xx)% of the defined structure: Structure Dose-Volume Parameter Bladder Rectum Femoral Head(s) D 50 D 35 D 25 D 15 D 35 D 25 D 20 D 15 D 40 D 20 D 5 Prostate D 95 Seminal Vesicles D 95 Table 5-1: Daily Dose-Volume Parameters 27

44 Normalized Volume (%) Normalized Volume (%) Prostate 1, Primary Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-1: "Prostate 1", Daily DVH vs. Planned DVH, Bladder, Primary Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-2: "Prostate 1", Daily DVH vs. Planned DVH, Rectum, Primary 28

45 Normalized Volume (%) Normalized Volume (%) Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-3: "Prostate 1", Daily DVH vs. Planned DVH, LT Femoral Head, Primary Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-4: "Prostate 1", Daily DVH vs. Planned DVH, RT Femoral Head, Primary 29

46 Normalized Volume (%) Normalized Volume (%) Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-5: "Prostate 1", Daily DVH vs. Planned DVH, Prostate, Primary Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-6: "Prostate 1", Daily DVH vs. Planned DVH, Seminal Vesicles, Primary 30

47 Normalized Volume (%) Normalized Volume (%) Prostate 1, Boost Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-7: "Prostate 1", Daily DVH vs. Planned DVH, Bladder, Boost Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-8: "Prostate 1", Daily DVH vs. Planned DVH, Rectum, Boost 31

48 Normalized volume (%) Normalized Volume (%) Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-9: "Prostate 1", Daily DVH vs. Planned DVH, LT Femoral Head, Boost Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-10: "Prostate 1", Daily DVH vs. Planned DVH, RT Femoral Head, Boost 32

49 Normalized Volume (%) Normalized Volume (%) Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-11: "Prostate 1", Daily DVH vs. Planned DVH, Prostate, Boost Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-12: "Prostate 1", Daily DVH vs. Planned DVH, Seminal Vesicles, Boost 33

50 5.1.3 Prostate 1, Average Daily Deviation Structure Dose-Volume Parameter Average % Difference from Planned Value (Primary, Daily) Average % Difference from Planned Value (Boost, Daily) D % 22.06% Bladder D % 24.87% D % 20.80% D % 10.91% D % 29.76% Rectum D % 18.87% D % 14.15% D % 9.27% D % -1.24% LT Femoral Head D % -2.60% D % -5.61% D % 0.19% RT Femoral Head D % -0.53% D % -0.33% Prostate D % -0.42% Seminal Vesicles D % (30.79%) Table 5-2: Prostate 1, Daily Average Percent Difference from Plan, Primary As a result of re-calculating the planned treatment fields to an isocenter equivalent to the treatment day soft-tissue 3D-3D match performed at the treatment console, the average percent difference between the expected and observed bladder, rectum, femoral head(s), and prostate dose-volume parameters was calculated. The primary prescription for Prostate 1 was 180 cgy per fraction for 25 fractions, followed by a boost prescription of 180 cgy per fraction for 18 fractions. The margins used for the primary prescription s PTV for Prostate 1 were 1cm in all directions, with a 0.7cm posterior margin as an expansion from the CTV, including the prostate and seminal vesicles. The margins used for the boost prescription s PTV for 34

51 Prostate 1 were reduced to 0.7cm in all directions, with a 0.5cm posterior margin as an expansion from the CTV, including only the prostate. The daily DVH to the seminal vesicles and the average percent difference are included above for completeness, although the seminal vesicles are not included in the PTV for the boost prescription. The complete set of daily percent difference for the above listed critical structures and dosevolume parameters are included in Tables A.1 Through A.4 of the appendix. Outliers observed in the primary dose of radiation included the rectum and seminal vesicles Day1 dose, RT femoral head Day 3 dose, and prostate Day 6 dose. The day one dose differences may be due to the Day 1 setup error, where the Day 3 and Day 6 outliers for RT femoral head and prostate correspond to two of the three highest complete shift vector length (CSVL) values observed over the primary dose of radiation, equaling 1.40 cm and 1.10 cm, respectively. This indicates the possibility of differences in dose due to setup error in the case of these outliers. In both the primary and boost, the dose-parameters examined for bladder and rectum were observed to be higher each day than what was planned. This may be due to differences in the method by which the rectum and bladder were contoured, however the consistent variability in the dose delivered to the bladder and rectum would indicate that the true cause of this variability is due to daily variations in the volume of bladder and rectum. 35

52 Normalied Volume (%) Normalized Volume (%) Prostate Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-13: "Prostate 2", Daily DVH vs. Planned DVH, Bladder Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-14: "Prostate 2", Daily DVH vs. Planned DVH, Rectum 36

53 Normalized Dose (%) Normalized Volume (%) Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-15: "Prostate 2", Daily DVH vs. Planned DVH, LT Femoral Head Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-16: "Prostate 2", Daily DVH vs. Planned DVH, RT Femoral Head 37

54 Normalized Volume (%) Daily DVH Planned DVH Absolute Dose (cgy) Fig 5-17: "Prostate 2", Daily DVH vs. Planned DVH, Prostate Prostate 2, Average Daily Deviation Structure Dose-Volume Parameter Average % Difference from Planned Value (Daily) D % Bladder D % D % D % D % Rectum D % D % D % D % LT Femoral Head D % D % D % RT Femoral Head D % D % Prostate D % Table 5-3: Prostate 2, Daily Average Percent Difference from Plan 38

55 As a result of re-calculating the planned treatment fields to an isocenter equivalent to the treatment day soft-tissue 3D-3D match performed at the treatment console, the average percent difference between the expected and observed bladder, rectum, femoral head(s), and prostate dose-volume parameters was calculated. The prescription for Prostate 2 was 250 cgy per fraction for 28 fractions with no boost. The margins used for the prescription s PTV for Prostate 2 were non-standard margins defined by an expansion of 5 millimeters in all directions from the CTV, with no expansion at the posterior aspect of the CTV. The complete set of daily percent difference for the above listed critical structures and dose-volume parameters are included in Tables A.5 and A.6 of the appendix. 5.2 Accumulated DVH vs. Planned DVH The accumulated doses for two prostate patients, Prostate 1 and Prostate 2 were evaluated using the commercially available deformable registration software, MIM version The Dose Accumulation - Deformable workflow was used to accumulate dose on daily CBCT images. For Prostate 1, a boost dose of radiation was prescribed. For this patient, the planned primary and boost dose calculation was compared with the MIM accumulated dose. In addition to these prescriptions being evaluated separately, a composite MIM accumulated dose was compared with the planned composite dose for the course of treatment. For Prostate 2, no boost dose of radiation was prescribed. For this patient, the planned DVH was compared with the MIM accumulated distribution. The evaluation of the total dose delivered to critical structures for each of these patients was done by comparing dose-volume constraints for femoral 39