AD_________________ (Leave blank) Award Number: W81XWH-07-1-0083 TITLE: On-line Adaptive Radiation Treatment of Prostate Cancer PRINCIPAL INVESTIGATOR: Tiezhi Zhang CONTRACTING ORGANIZATION: William Beaumont Hospital Royal Oak, Michigan REPORT DATE: January 2009 TYPE OF REPORT: Annual PREPARED FOR: U.S. Army Medical Research and Materiel Command Fort Detrick, Maryland 21702-5012 DISTRIBUTION STATEMENT: (Check one) √ Approved for public release; distribution unlimited Distribution limited to U.S. Government agencies only; report contains proprietary information The views, opinions and/or findings contained in this report are those of the author(s) and should not be construed as an official Department of the Army position, policy or decision unless so designated by other documentation. Form Approved REPORT DOCUMENTATION PAGE OMB No. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202- 4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. 1. REPORT DATE 2. REPORT TYPE 3. DATES COVERED 31-01-2009 Annual 1 JAN 2007 - 31 DEC 2008 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER On-line Adaptive Radiation Treatment of Prostate Cancer W81XWH-07-1-0083 5b. GRANT NUMBER 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER Tiezhi Zhang 5e. TASK NUMBER 5f. WORK UNIT NUMBER Email: 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER William Beaumont Hospital Royal Oak, Michigan 9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S) U.S. Army Medical Research and Materiel Command Fort Detrick, Maryland 21702-5012 11. SPONSOR/MONITOR’S REPORT NUMBER(S) 12. DISTRIBUTION / AVAILABILITY STATEMENT Approved for Public Release; Distribution Unlimited 13. SUPPLEMENTARY NOTES 14. ABSTRACT None provided. 15. SUBJECT TERMS None provided. 16. SECURITY CLASSIFICATION OF: 17. LIMITATION 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON OF ABSTRACT OF PAGES USAMRMC a. REPORT b. ABSTRACT c. THIS PAGE 22 19b. TELEPHONE NUMBER (include area U U U UU code) Table of Contents Page Introduction…………………………………………………………….………..….. 3 Body………………………………………………………………………………….. 3-5 Key Research Accomplishments………………………………………….…….. 6 Reportable Outcomes……………………………………………………………… 6-7 Conclusion…………………………………………………………………………… 7 References……………………………………………………………………………. 7 Appendices…………………………………………………………………………… 8 2 RE: DOD Grant W81XWH-07-1-0083 PROGRESS REPORT SUMMARY INTRODUCTION The goal of this project is to develop an online adaptive treatment technique for prostate cancer treatments. During the first year, we have developed parallel deformable image registration, parallel dose calculation and plan optimization algorithms. During the second year, we have been focusing on improving the robustness of the algorithms and improving online image quality. STUDIES AND RESULTS: YEAR 2 A. Improving image quality Image quality of cone beam CT (CBCT) is a major obstacle in clinical implementation of online adaptive treatment. CBCT images lack of contrast, and therefore it is not possible to delineate some of the region-of-interest (ROI). The main reasons for poor image quality are due to excessive scatter and suboptimal detector performance. We have designed a new online imaging system – Tetrahedron Beam Computed Tomography (TBCT) that can conquer these problems1,2. Figure 1 The new system design for online imaging: Tetrahedron Beam Computed Tomography (TBCT): (a) mounted on radiotherapy treatment machine, (b) diagram of scanning geometry. Figure 1 shows the design of TBCT. This system uses a linear scan x-ray tube and a linear detector array. Due to its unique geometry, the majority of scatter photons are rejected. In addition, it uses the same type of high quality detector as diagnostic CT scanners. Therefore, it is expected to be able to achieve the same image quality as a diagnostic CT scanner. We have acquired an NIH R21 grant to develop a benchtop system. This project was also presented at the 2008 AAPM annual conference. A manuscript has been submitted to Medical Physics Journal, and is currently under review. 3 B. Development of a fast treatment delivery method During the first year, we implemented a fast parallel treatment plan optimization method using a Beowulf cluster with a Message Passing Interface. To prepare for clinical implementation, in the second year, we are focusing on accelerating treatment delivery. While online adaptive planning takes care of interfraction motion, the target may also undergo significant intrafraction motion due to patient and bowel movement. It is also noted that intrafraction motion increases with treatment time. This justifies the importance of a faster treatment delivery method. VMAT is a new dynamic treatment delivery method which can significantly shorten the treatment time. The treatment planning method for VMAT however is not mature. We are developing a robust VMAT treatment planning method which incorporates MLC constraints into the optimization process. Figure 2 shows the flow chart of our IMAT treatment planning method. It starts with regular sliding window IMRT treatment planning then corrects the rotation angle difference in iterations. One major advantage of this treatment planning method is the robustness. This is a very important feature for online adaptive treatment since a new plan is generated for every treatment. Figure 3 compares the difference between the Figure 2 Flow chart of IMAT treatment IMAT treatment planning method we developed planning based on iterative fluence intensity and regular 7 field IMRT treatment planning. optimization and sliding window conversion. We can see the dose distribution from IMAT treatment planning is very similar to regular IMRT treatment. It is noted that the target dose by IMAT treatment planning is more uniform than IMRT. The overall objective score of IMAT is lower than that from 7 field IMRT (0.0699 vs. 0.0994, respectively). We believe the slight improvement is due to the freedom of selecting optimal angles in IMAT planning. The major improvement in the IMAT method is that it can be delivered within 2-4 minutes, compared to 15 minutes for a regular IMRT treatment. While primarily developed for prostate online adaptive treatment, the new IMAT treatment planning method shows much broader applications in all treatment sites. Head and neck VMAT treatment planning was difficult because of the difficulty of converting complicated fluences to deliverable MLC leaf sequences. Our new method solves this problem by taking account of MLC constraints during optimization iterations. Because of the freedom of choosing gantry angles, the IMAT treatment planning is more computationally expensive. This problem can be easily solved by using the parallel dose calculation and optimization algorithms that we have developed previously. Besides that, we are also working on Graphic Processing Unit (GPU)-based parallel computing for dose calculation and optimization. The GPU is a cost effective parallel computing resource. A single GPU may 4 surpass the computing capacity of a medium scale computer cluster. We expect application of the GPU will further boost the speed of computation for dose calculation and optimization. IMRT IMAT 1 0.9 0.8 DMPO:Bladder 0.7 DMPO:Prostate DMPO:Rectum e 0.6 m volu 0.5 DIMMAPTO:B:Rlaidndger % 0.4 IMAT:Prostate 0.3 IMAT:Rectum 0.2 IMAT:Ring 0.1 0 0 50 100 150 200 250 Dose (cGy) Figure 3 Comparison of IMRT treatment planning (top) and IMAT treatment planning (bottom). The dose distributions and DVHs are very similar, but IMAT treatment can be delivered much faster, so the impact of intrafraction motion will be minimized. We will submit abstracts to AAPM and ASTRO annual conferences in March 2009 regarding to the IMAT treatment planning method. We also plan to submit the implementation of GPU-based parallel optimization and deformable image registration to AAPM. Manuscripts of publication will be prepared and submitted a few months later. C. Model-based automatic segmentation of ROI for prostate treatment Previously, we developed intensity-based deformable image registration algorithms3,4. The algorithms have been proven to be very robust in some treatment sites, such as head and neck, breast and lung. However this algorithm is not often successful for organs in the pelvis. This is due to the extremely large deformation of the bladder and large variation of the contents in the rectum. To solve this problem, we decided to change our plan. We are working on combining the model- based automatic segmentation with intensity based segmentation. We represented the organ using an ROI surface mesh. Figure 4 shows the surface mesh of the rectum. The surface mesh 5 is created on treatment planning CT images, then distance transformations are performed to calculate the distance of pixels to the boundary. With boundary and distance information, we can treat voxels inside and outside organ differently. We can also weight the voxels basing on its distance to the boundary. So we can emphasize image information closer to the boundary. For the shape change, we are currently using penalties on nodal displacement. When we accumulate enough data, we can perform PCA analysis to determine the eigenspace that organs deform and penalize the deformation that is out of the eigenspace. Figure 4 Surface mesh of the rectum. Using the surface representation, we can use distance information to emphasize on voxels closer to the boundary. We are also able to limit the shape change based on statistical model. KEY RESEACH ACCOMPLISHMENT 1. We have invented a new imaging system which can produce diagnostic quality images online for image guided radiotherapy. We are building a benchtop system with an NIH grant. We expect a clinical prototype system will be built within three years. 2. We have developed an IMAT treatment planning method that significantly shortens treatment delivery time. This method can minimize the impact of intrafraction organ motion during online adaptive treatments. We expect clinical implementation of this treatment planning method within 1-2 years. 3. We are currently combining voxel intensity-based deformable image registration with model based image segmentation. This method may overcome the problems of inconsistent rectum filling and large organ deformation. We expect this method can be fully developed within a few months, after which large scale evaluations will be performed. REPORTABLE OUTCOMES: Peer reviewed publications: 1. Tiezhi Zhang, Derek Schulze, Xiaochao Xu, “Tetrahedron Beam CT (TBCT): A New Design of Online Imaging System for Image Guided Radiotherapy”, Submitted to Medical Physics Journal. 2. Tiezhi Zhang, Yuwei Chi et al, “Automatic Delineation of Online Head-And-Neck Computed Tomography Images: Toward Online Adaptive Radiotherapy”, International Journal of Radiation Oncology, Biology and Physics 2007 68(5): 1572-8. 3. Derek Schulze, Tiezhi Zhang, “Comparison of various online IGRT strategies: The benefits of online treatment plan re-optimization”, Radiotherapy and Oncology, in press. 6 Conference presentations: 1. T. Zhang, D. Schulze, “Tetrahedron Beam CT (TBCT): A New Design of Online Imaging System for IGRT”, AAPM annual conference, 2008. 2. D. Schulze, T. Zhang, “Techniques of Online IMRT Plan Re-Optimization for Prostate Cancer Treatments”, 50th ASTRO annual conference, Boston, MA, 2008. 3. T. Zhang, et al, “Online delineation of head and neck CT images”, 49th AAPM annual conference, Minneapolis, MN, 2007 4. T. Zhang, D. Schulze, et al, “Online Adaptive Prostate Cancer Intensity Modulated Radiation Treatment (IMRT): Method of Online Plan Re-optimization” 49th ASTRO annual conference, Los Angels, CA, 2007 5. D. Schulze, T. Zhang, et al, “Dosimetric Comparison of Various Online Adaptive Prostate Cancer Treatment Techniques”, Los Angels, CA, 2007 6. T. Zhang, et al, “Clinical Applications of 3D and 4D Deformable Image Registration for Image Guided Radiotherapy”, 48th AAPM annual conference, Orlando, FL, 2006 7. T. Zhang, et al, “Automatic Delineation of Daily CT Images for Online Plan Adjustments: Method and Quantitative Validation”, 48th ASTRO annual conference, Philadelphia, PA, 2006 8. L. Burgess, T. Zhang, et al, “Image Guided Radiotherapy by Online Plan Re- optimization: Studies of Dosimetric Benefits by Treatment Simulations”, 48th ASTRO annual conference, Philadelphia, PA, 2006 Degrees obtained: 1. Derek Schulze, Master of Science, Department of Medical Physics, Wayne State University, Detroit, Michigan Funding applied for: 1. Development of A Quasi-CBCT System for Image Guided Radiotherapy, PI, Tiezhi Zhang, NIH R21, Started on 3/1/2008 In this proposal, we will develop a novel imaging system using a linear x-ray source and a linear detector. This imaging system may significantly improve the quality of online images, which is critically important for online ROI delineation. CONCLUSIONS: During the second year of this study, we are aiming at future clinical implementation of online adaptive treatment. We have invented a new imaging technique (TBCT) to improve online image quality. We have also developed an IMAT treatment planning method which can shorten the treatment time, thereby minimizing intrafraction motion. Automatic image segmentation of ROIs for prostate treatment remains a very challenging issue. We are developing a model-based segmentation method which may overcome the existing problems. We expect this method can be developed within a few months. We will perform a large-scale evaluation study thereafter. REFERENCE: 1. T. Zhang, D. Schulze, “Tetrahedron Beam CT (TBCT): A New Design of Online Imaging System for IGRT”, AAPM annual conference, 2008 (Abstract). 2. T. Zhang, Derek Schulze, Xiaochao Xu, “Tetrahedron Beam CT (TBCT): A New Design of Online Imaging System for Image Guided Radiotherapy”, Submitted to Medical Physics Journal. 7 3. T. Zhang, Y Chi et al, “Automatic Delineation of Online Head-And-Neck Computed Tomography Images: Toward Online Adaptive Radiotherapy”, International Journal of Radiation Oncology, Biology and Physics 2007 68(5): 1572-8. 4. Y Chi, J Liang, T Zhang, et al, “Automatic Contour Delineation On Cone Beam CT (CBCT) and Verification”, American Association of Physics in Medicine 50th Annual Conference, Huston, TX, 2008 (Abstract). 5. T Zhang, et al, “Online delineation of head and neck CT images”, 49th AAPM annual conference, Minneapolis, MN, 2007 (Abstract). 8 AAPM annual conference 2008 abstract: Title: Tetrahedron Beam CT (TBCT): A New Design of Online Imaging System for IGRT Authors: Tiezhi Zhang, Derek Schulze Purpose: Cone-beam computed tomography (CBCT) is an important online imaging modality for image-guided radiotherapy (IGRT) as well as other forms of image guided interventions. However, current CBCT image quality is inferior to that of the diagnostic fan beam CT. We have designed a novel Tetrahedron Beam Computed Tomography (TBCT) imaging system that may achieve the same diagnostic quality as helical CT scanners. Material and Methods: The TBCT imaging system is comprised of a linear scan x-ray source and a linear discrete x-ray detector array. The axis of linear x-ray tube and the detector array are aligned perpendicular to and within the rotation plane, respectively. The x-ray beams are narrowly collimated into fan beams and focused to the linear detector. Detector and x-ray tube rotate slowly while the fan beams scan quickly along the axis. The TBCT reconstruction geometry is similar to CBCT. Approximate and exact reconstruction algorithms can be modified for TBCT reconstruction. Results: TBCT will produce diagnostic quality online images due to its scatter rejection mechanism and the use of high-performance discrete x-ray detectors. TBCT also has several other advantages such as larger clearance, ease of performing dynamic field size and mAs controls, etc. Conclusion: TBCT will significantly improve online image quality. Clinical implementation of TBCT would be of importance in IGRT as well as other forms of image guided interventions. .