There has been a widespread increase in the use of radiotherapy, especially intensity modulated
radiation therapy (IMRT), an advanced form of external beam irradiation, for the management of
prostate cancer that shows significant potential for improving the therapeutic ratio since radiation
can be shaped to cover the prostate while avoiding healthy nearby tissue. The current gold standard
is to use implanted metallic prostate markers and kV X-ray based images for daily verification of
position. Therefore, to ensure accurate prostate placement throughout the treatment, real-time 3D
tracking of the prostate is required. Dr. Wiersma investigated several prototypes for the motorized
kV collimator before arriving at the current design that meets the necessary constraints and has
stable performance. He is in the process of characterizing the kV aperture. Additionally, the IR
tracking device provides accurate real time 3D positional data. The imaging system will be interfaced
from Labview allowing synchronization of the output positions with the motor control inputs.

Currently, kV imaging for patient setup leads to additional dose over and above the prescribed
IMRT treatment dose. Inclusion of the kV dose at the treatment planning stage would allow the
benefit of high contrast kV imaging without this additional dose overhead. A Monte Carlo scheme
was developed to model dose deposition kernels that were imported into a commercial treatment
planning system. This will allows users the ability for combined MV and kV dose planning and more
targeted therapy delivery. For example, while the cone beam computed tomography (CBCT) delivers
dose to all of a patient’s skin, the reduced aperture MV-kV technique spares most of the skin. Most
importantly, the diagnostic dose to the tumor is substantially higher than the diagnostic dose to
healthy tissue, since all of the kV beams intersect at the point of treatment.

Finally, Dr. Wiersma has developed in-house software using the C# programming language to
automatically segment the fiducials from the surrounding image data. He has also made preliminary
progress on developing a method to optimize the size of the kV aperture while still allowing full
imaging of the markers.

Aim 1. Design and develop a motorized kV collimator: over the period of grant funding, we have investigated several prototypes before arriving at a design that meets the necessary constraints and has stable performance (Fig. 1c). The most recent prototype consists of an aluminum backing plate machined to fit in place of the standard bowtie filter. To reduce the size and mass of the collimator, the body was made using a thin magnesium sheet. Low friction linear glide rails hold four interchangeable jaws that can be independently moved using linear servos. Control electronics have been extensively upgraded. Most notable is replacement of the external computer by a small onboard micro-computer (Fig. 2). We are in the process of characterizing the kV aperture. For spatial and temporal accuracies two different measurement devices are used; (1) a linear encoder attached to the collimator jaw, and (2) a stereoscopic infrared (IR) imaging system to track markers attached to the jaws. An encoder allows very accurate position measurements with very fast response time. Additionally, the IR tracking device provides accurate real time 3D positional data with an approximate 0.25mm spatial resolution at a 60 Hz measurement rate. The imaging system will be interfaced from Labview allowing synchronization of the output positions with the motor control inputs. We anticipate that the results will be submitted as an articled titled,“Development of a Dynamic kV Collimator for Low Diagnostic Dose RealTime 3D Motion Tracking during Radiation Therapy by Combined MV-kV Imaging” to the IOP journal of Physics in Medicine, and Biology by the end of 2011. Figure 5: DVH for a reduced 3×3 cm2 aperture (solid) and full 20×20 cm2 field (dashed). Figure 3: Comparison of measured (diamond detector) and calculated (Pinnacle) precent depth dose in water for a 125 kVp 10cm x 10cm beam. Figure 4: Left, dose for 20×20 cm sized aperture. Most of the area is covered in dose equal to the top half of the dose range for kV imaging. Right, dose distribution for reduced (3×3 cm) aperture. Here dose falls in the lower half of the dynamic range of dose from kV imaging. Aim 2. Incorporation of the kV dose at the treatment planning stage: all objectives outlined in Aim 2 were accomplished. A Monte Carlo scheme was developed to model dose deposition kernels that were imported into a commercial treatment planning system (Pinnacle, Philips Medical Systems, Madison, WI). The kV source was commissioned in the treatment planning system using water tank measurements with a diamond detector (Fig. 3). For comparison purposes, a typical CT procedure for patients receiving radiation therapy for prostate cancer is ~ 6 cGy. This was found to be comparable to the dose delivered by a 3×3 cm aperture in fluoroscopic mode for MV-kV imaging (Fig. 4). However, while the CBCT delivers dose to all of the patient’s skin, the reduced aperture MV-kV technique spares most of the skin, as seen in Fig. 4. Most importantly, the diagnostic dose to the tumor is substantially higher than the diagnostic dose to healthy tissue, since all of the kV beams intersect at the point of treatment. A dose volume histogram (DVH) is calculated to provide a quantitative measure of the tissue sparing effect of the reduced aperture. Fig. 5 plots DVH curves for both 20×20 cm and 3×3 cm kV apertures. As can be seen, the volume of tissue (cm3 ) exposed to radiation is significantly reduced for the 3×3 cm field. Figure 6: Screen capture of in house software tracking one of three gold implanted markers (red box) embedded in a Rando pelvic phantom. Aim 3. Development of an algorithm that will use prediction to calculate optimally shaped kV apertures: in-house software was developed using the C# programming language to automatically segment the fiducials from the surrounding image data (Fig. 6). Progress on developing a method to optimize the size of the kV aperture while still allowing full imaging of the markers has been preliminary. Currently, we have included a tool in the software that allows the user to manually set the X1, X2, Y1, Y2 jaw sizes. We anticipate that completion of the kV aperture characterization (Aim 1) will allow sufficient mechanical constraint data (max. jaw velocity, jaw travel extent) to program an aperture optimization algorithm. In addition, our proposed follow up work to Aim 2 would naturally calculate the most optimal kV aperture. Future Outlook: Progress on the design, construction, and implementation of a dynamic kV collimator for low dose marker tracking has proceeded well. We anticipate submission of our data for publication by the end of 2011. Additionally, we plan to contact the University of Chicago Office of Technology and Innovation to submit a public disclosure on the method. As follow up work, we have initiated a new project that aims to combine the concepts learned from Aims 2 and 3 into a unified approach. The PI is currently mentoring a graduate student, Zachary Grelewicz, who is presently supported under the CRF grant, on carrying out such research. Again, the PI would like to thank the Cancer Research Foundation for supporting his research at an early stage in his academic career. This has allowed sufficient gathering of preliminary data necessary for successful future funding. We anticipate submitting grant proposals in 2012 to the NIH, DoD, and ACS based on these results. FINAL PROGRESS REPORT Aim 1 has been completed as outlined in the grant application. A functional prototype kV collimator has been built and tested. We are in the final stages of securing a Master Research Agreement with Varian Medical Systems (Palo Alto, CA) that will provide research funding to investigate projects approved by Varian. One proposed project, lead by my colleague, Charles Pelizzari, PhD will use the collimator developed in this grant to investigate dynamic region-ofinterest imaging for dose reduction and image quality improvement in IGRT. Here we are exploring the use of the kV collimator for use in reduced region-of-interest cone beam CT imaging. Currently, existing CBCT acquisition and reconstruction methods require that the entire patient cross section be illuminated by the X-ray beam, even if an image of only a small region of the entire cross-sectional image is really required for the clinical application at hand. This results in unnecessary dose to tissues outside the region of interest, and additional contamination of the images by photons scattered from those unnecessarily irradiated tissues. As stated in my previous progress report, we have combined Aims 1 and 2 into a new research project titled, A Hybrid Imaging and Therapy Optimization Approach for Reducing Treatment Delivery Errors during Prostate Radiation Therapy. This project is currently under review for funding as a New Investigator Prostate Idea grant under the Department of Defense Congressionally Directed Medical Research Programs (CDMRP) and as a research grant under the American Cancer Society Illinois. We hope by simultaneous optimization of the MV treatment beam through the multi-leaf collimator (MLC) and the kV beam through the dynamic aperture, that the kV imaging radiation can be effectively merged into the MV treatment planning dose as to allow real-time 3D monitoring of the prostate position during delivery for the first time. This has potential to further increase the therapeutic effectiveness of RT and decrease normal tissue complications by allowing accurate and safe delivery for both conventional and hypofractioned prostate RT. The proposed research will consist of the following primary aims: (1) Development of a dual MV+kV dose/imaging optimization engine using PTV, OAR, and image based constraint parameters. (2) Design, construction, and optimization of a dynamic aperture for kV beam modulation. (3) Full tracking and dosimetric method validation using a 3D motion phantom, TLDs, and data from prior treated prostate patients. Successful completion of the proposed work would allow real-time 3D prostate motion monitoring to be cost effectively implemented in general clinics around the world using standard LINAC equipment for the first time. With an estimated 241,740 new cases of prostate cancer in the US for 2012 alone, and with RT being one of the primary treatment modalities, easy access to real-time prostate motion tracking would be of considerable impact in terms of ensuring better therapeutic outcomes precise targeting, preventing normal tissue complications, allowing use of newer treatment options as hypofractionation which have potential for higher curative rates, and removal of unwanted imaging dose overhead currently associated with existing kV imagine guidance techniques.