Events Calendar
Abstract: The quest for very fast, within clinical turn-around times and accuracy, proton Monte Carlos has been realized with various commercial and research systems. Our internal clinical effort started in 2014 with the (now Dr.) Depauw’s thesis. In it, he described and implemented a proton Monte Carlo with the aim to reduce the problem to the most minimal, and presumably highest performance, physics description of the problem. This resulted in a model where proton scatter and transport was reduced to a per voxel scatter at an energy-material specific constant angle, a random azimuthal angle (the only random angle in the transport!) and energy loss per track derived from measured and clinically validated depth doses. We, especially, deem the latter an important GMC feature as it obviates the need for tuning required to match MC results to measurements. This produces a transparent link between GMC dosimetry and reference dosimetry and workflows. The thesis implementation proved the applicability of the model in terms of performance on a PC and achieved dosimetry. The GPU implementation (Windows 10, RTX5000) has added many clinically necessary features, including apertures and shifters, secondary scatter correction, and other clinical necessities. The latter includes computation of voxel dose-averaged LET and post-processing for Unkelbach bio-dose (our term) and (in progress) RBE based on organ-specific a/b in the McNamara model. Performance is very high and plans are achieved on the order of 5 sec (with 50 106 protons / sec). GMC input is a DICOM set including CT, RT Ion Plan, RT Struct and RT Dose as well as a compute context to indicate the type of computation (default PLANLEVEL RT Dose); no dependency exists on the treatment planning system. We will deploy GMC on a departmental GPU server (fault-tolerant). Communication with the server is provided through a management layer, ASAP, which communicates with Whiteboard or through user commands. Whiteboard connectivity will integrate GMC with automated workflows. We expect GMC, with TOPAS as a gold standard, to manage the transition from pencil-beam dose prescriptions to MC dose prescriptions.
About the speaker: Arrived in the US for graduate study in High Energy Physics in 1976 from Delft, The Netherlands. Obtained PhD in HEP working at the Cornell Synchrotron (10 GeV com). Worked one year for Xerox in Rochester NY where I switched to Radiation Oncology / Medical Physics. Was invited to join the Joint Center for Radiation Therapy in 1987. Responsible for Treatment Planning Operations and Stereotactic Radiosurgery and Radiotherapy (Loeffler, Tarbell & Alexander). Joined MGH around 1999. Primary focus on engineering and clinical deployment of Burr. Interests in system and computational approaches to problems in radiotherapy.