SU‐FF‐T‐159: Inverse Planning for Segment‐Modulated Arc Therapy (SMART) as a 0–1 Labeling Problem

Purpose: Segment‐modulated arc therapy (SMART) is an efficient and precise radiotherapy modality which is recently available (e.g. Varian RapidArc). The purpose of this work is to propose a novel algorithm for SMART inverse planning by using the concept of binary image labeling. Method: The aim of SMART treatment planning is to find an optimal segment shape and weight for each of a series of consecutive control points to achieve a clinically favorable dose distribution. For this purpose, the planning objective is defined as the sum of a traditional dose/volume objective and a segment objective. By conceiving segmented field as a binary image (1 for open, 0 for close) defined on the MLC plane, the segment objective is set as the total‐variation of the image along the leaf‐motion direction. In the proposed algorithm, the segment shape is not managed by its boundary. Instead, a segment is directly modulated by opening or closing the pixels. To optimize the system, simulated annealing (SA) technique is utilized. Initially, a segment is set by randomly open/close the image pixels. In each step of SA, a beamlet is randomly selected and switched (0→1 or 1→0). The segment weights are determined applying quadratic programming to minimize the dose/volume objective. Then the segment objective and the total objective are calculated. SA algorithm stops when the planning criteria are satisfied. Results: The proposed algorithm was evaluated by a prostate case and it is observed that our algorithm produces a highly conformal IMRT plan. By applying the segment objective, the open pixels in a segment are gradually connected in leaf motion direction as the system converges and a deliverable segment forms. Conclusions: A novel SMART inverse planning algorithm, working on beamlet space by using binary labeling concept, was developed and tested.