Development and in silico validation of an online adaptive workflow for proton therapy with an in-room synchrocyclotron and CT-on-rails.

Abstract

Development and validation of an online adaptive proton therapy workflow for clinical use. A script-based online adaptive proton therapy workflow was developed in a graphics processing unit (GPU)-accelerated treatment planning system (TPS) to automate plan adaptation using traditional CT images acquired on an in-room CT-on-rails. Monte Carlo (MC) -based secondary dose calculation and log file analysis of machine parameters were integrated into the workflow to provide quality assurance (QA) pre- and post- plan delivery for the adapted treatment plan. In silico retrospective testing was performed with ten previously treated pelvic patients for validation of the workflow, a total of 50 fractions were included in this study. The fidelity of adapted dose distributions, time required for the in-silico workflow, and QA results were recorded and analyzed. The median (inter-quartile range, IQR) time required for this workflow was 61.9 (12.2) minutes, demonstrating technical feasibility for online adaptation. Of the 50 evaluated fractions, 15 (30%) showed clinically meaningful improvements in sparing of highest-priority organ-at-risk (OAR) using the adaptive plans (P_A), and 8 (16%) showed improved target coverage relative to the scheduled plan. Both MC-based secondary dose checks and log file-based machine QA were within institutional tolerance criteria, confirming the dosimetric accuracy and delivery fidelity of the workflow. A fully integrated, QA-embedded online adaptive proton therapy workflow using in-room CT-on-rails imaging was developed and validated in silico on a clinical proton beamline. The workflow demonstrated clinically acceptable treatment times and reliable dosimetric accuracy and provided measurable benefits in OAR sparing and target coverage for a substantial subset of fractions. These results support the readiness of this workflow for clinical use.