Session Item

Treatment plan evaluation based on the impact of anatomical variability often uses images acquired during the course of treatment (e.g., weekly CT), therefore can only inform the planning process for a portion of the treatment delivery. We suggest including systematic progressive changes (SPCs) of a patient population into plan evaluation to provide additional information before treatment. 

20 radiotherapy patients with nasopharyngeal cancer were included in this retrospective study. Each patient had a planning CT and 6 weekly CTs during treatment. To build the anatomical model, we deformed the weekly CTs to the planning CTs of our training population (n=19) and obtained the average anatomical change per week. To predict a deformation for the remaining patient, the average deformation of the training population was applied to the patient’s planning CT presenting SPCs. K-fold cross-validation (n=4) was used to obtain a sample of 4 patients. For those 4 patients, IMPT plans using 3-field (3B), 4-field (4B), and 5-field (5B) beam arrangements were generated. Each IMPT plan was forward calculated on the plan evaluation scenario and the dose was accumulated (AD). We compared two before-treatment evaluation methods: 1) conventional robust evaluation using 3 mm setup uncertainty; 2) SPC evaluation using predicted weekly CTs with 3 mm shifts. We ranked the IMPT plans based on the AD for each considered dose metric. A lower AD gave a higher rank for organs at risk, while a higher AD gave a higher rank for the CTV. The gold standard rank was established from the AD based on weekly CTs. The rankings from the before-treatment evaluations are compared to the gold standard ranking to quantify the performance of the evaluation method.

Figure 1 shows the dose distributions of the three plans and corresponding DVH curves (full and CTV zoom) for the example patient, ranking is displayed in table 1. In terms of CTV coverage, we observe that SPC evaluation agrees with the gold standard and ranks the 3-field plan most robust against anatomical deformations, whereas conventional evaluation would rank the 5-beam plan most robust. For CTV and OARs, DVHs of the SPC evaluation agree better with the DVHs of the gold standard evaluation compared to conventional evaluation, demonstrating that the SPC evaluation better captures the patient’s deformations and associated dose degradations. Over all patients, among the total of 48 ranking comparisons (12 considered dose metrics per patient), we discovered that SPC evaluation was better or equal than conventional evaluation in 44/48 of cases. Only in 4/48 cases, SPC evaluation is inferior to conventional evaluation.

Including the influence of SPC during robust evaluation can assist in plan evaluation and decision making, providing information on clinical uncertainties in addition to the conventional rigid isocenter shift and CT number uncertainty. This can facilitate robust treatment plan selection for IMPT for head and neck cancer.