Implementation of an Automatic-Planning System for Breast Cancer Radiotherapy Planning

Abstract

As a tertiary referral canter, we treat over 720 new breast cancer (BC) patients annually. BC-Radiation Therapy (RT) is done using forward “field-in-field “(FiF) planning which shown to improve dose homogeneity and reduce RT-related toxicity. The workload on the planning team (physicists/dosimetrist) exceeds the recommendation of the IAEA. As part of our efforts to improve and reduce the workload, we evaluated the use of a fully automatic-planning system (APS) (EZFluence) for planning these cases. In the current study we report our experience for the use of APS in the first 10 BC patients in the adjuvant setting. The study is part of a general IRB approval for BC-RT at our institution. CT-simulation and target volume delineation was done per department protocol. RT fractionation was 2.65 Gy X 16 fractions. All cases were planned twice by the breast planning team (only one attempt for each): tangential forward FiF planning and APS. Plan objectives were PTVeval ≥ 95% while maintaining organs at risk constraints. The time to initiate a RT plan, and radiation oncologist (RO) preference were recorded. Dosimetric parameters compared included conformity Index (CI) and Dmax. The APS was limited to optimize not more than 4 segments per field with a minimum of 4 MU per field. CI = volume of 95% of the prescribed dose divided by the target volume (TV), DMax represents the highest dose at 2cc of TV. The use of the APS reduced optimization time of FIF by 70% (52±8 minutes vs 15±3minutes, p≤0.00001). The CI was 1.41±0.14 for APS vs 1.49±0.18 for manual optimization (p = 0.017). Dmax was 104.3%±1.8 and 108.5±2 for APS vs. manual FiF, respectively (p≤0.0001). The number of segments for the manual FiF plan was between 1-4, median 2. RO preferred the APS plan in all 10 cases. Optimization time and dose homogeneity were in favor of APS for BC RT planning. Dose homogeneity (e.g. Dmax) was most probably the key parameter for RO preference of the APS plan. This approach potentially improves workflow (time per plan, number of planning attempts), freeing skilled dosimetrists to perform other tasks and reduced the workload at a high-volume RT unit.