A GPU-Based Monte Carlo QA Tool for IMRT and VMAT
Y Graves1*, G Kim1, M Folkerts1, T Teke2, I Popescu3, L Cervino1, Z Tian1, X Jia1, S Jiang1, (1) University of California, San Diego, La Jolla, CA, (2) BC Cancer Agency, Kelowna, BC, (3) British Columbia Cancer Agency, Vancouver, BCWE-E-BRB-9 Wednesday 2:00:00 PM - 3:50:00 PM Room: Ballroom B
Purpose: To develop a GPU-based Monte Carlo (MC) 3D dosimetry quality assurance (QA) tool employing patient geometry and actual delivery information.
Methods: First, we generate fluence maps at all beam angles from the initial treatment plan. A GPU-based MC dose engine, gDPM, is employed for the secondary dose calculation (SDC) on patient CT. This SDC is used to verify the TPS plan dose (PD) accuracy. Before the 1st treatment fraction, we deliver the treatment plan on a Linac without any phantom setup to obtain machine log files. With the log files, we extract actually delivered fluence maps at all beam angles and perform delivered dose calculation (DDC) using gDPM. The difference between DDC and SDC indicates possible errors in data transferring and machine delivery. Lastly, the comparison between DDC and PD shows the accumulative errors from all the possible sources. Moreover, a web application for this QA tool is developed for clinical use. We have tested this QA tool on 6 patients, 4 VMAT and 2 IMRT patients. We reported mean gamma values and passing rates inside the 20% isodose line; DVH plot and dose difference matrix are also documented.
Results: For all six patients, the gamma passing rates within the 20% isodose line for SDC, DDC and PD comparisons are all higher than 95%. In the DVH plot, the three dose distributions were found to be very close. A typical IMRT or VMAT case takes less than one minute to run the whole QA tool.
Conclusions: We have developed a GPU-based MC QA tool which can be used for efficient and easy IMRT and VMAT QA.