Context and objective With the arrival of hybrid Photon-MRI solutions in standard radiotherapy, more research are going towards real-time adaptive treatment or real-time treatment verification applications. In proton therapy, the hardware solution doesn’t exist yet but is under serious consideration. Proton treatments, with their particular dose distribution profile, should benefit even more than photons of such an hybrid device. In this context, we developed a tool to study the possible new treatment strategies that could emerge from such an hybrid machine and their possible gain. This particular work aims at visualizing a PBS treatment plan on the continuous 2D MR images and compare a real-time recomputed range (RR) based on those images with the planification range (PR) computed on the average CT. Method To compute the proton energy loss, stopping power information need to be extracted from CT images in order to achieve a reasonable accuracy. First, the MRI continuous acquisition is launched and the diaphragm position is tracked using an home-made tool. Based on the tracked position, the correct 4DCT slices and 4DCT phase corresponding to the current MRI slice is selected automatically. Then, a non-rigid registration between the current MRI slice and the selected 4DCT slice is performed to create a virtual continuous CT video. Finally, the RR is recomputed on this video, which imitates the real-time motion captured by the MRI better than the 4DCT. Results and conclusion Early results show a significative mean difference between the RR and PR, with specific spot difference from -9 to 54 mm (Mean = 20 mm, Std = 11 mm). The mean difference wasn’t affected much by delivery starting time for the studied patient. By taking motion into account, the developed tool revealed significant deviations between the planned range and the recomputed range. Such tool would therefore be very helpful to assess the robustness of the treatment plan to motion effects.
Dasnoy-Sumell, Damien ; et. al. Towards Real-Time proton range verification using dynamic MRI.4D treatment planning workshop (Vienna, du 04/12/2017 au 05/12/2017).