BLUE RIBBON POSTER THERAPY: Nopause: A Blinded, Multiple-Observer Analysis of Clinical Acceptability and Preference for Automated VMAT Plan Optimization.
Abstract
Purpose
Previous work suggests that automatically optimized plans were dosimetrically non-inferior to manual, interactively optimized plans based on institutional clinical goals in a retrospective paired analysis. This study aimed to extend this work to clinical acceptability and physician preference based on expert review.
Methods
The Northern Plan Automation Services project is a collection of in-house applications and scripts designed to improve efficiency and quality in radiotherapy treatment planning. The core platform is the Treatment Planning Automation System (TPAS) which automates the optimization of VMAT plans within our Eclipse/ARIA environment. TPAS was used to retrospectively generate treatment plans for 48 cases. Two radiation oncologists (ROs), one from outside our institution, compared the TPAS and manual plan and were asked to (a) score each plan on a 1-5 (5=ideal) Likert scale of clinical acceptability, (b) indicate which plan they preferred, and (c) indicate the strength of their preference on a scale of 1-3 (3=strong preference). Both ROs (referenced as “A” and “M”) were blinded to which plan was created by TPAS, as well as each other’s scores. Scores were analyzed using summary statistics and paired t-testing to determine if significant differences (p<0.05) were present.
Results
The mean clinical acceptability score for the manual plans was 4.5 (A) and 4.1 (M), compared to 4.7 (A) and 4.2 (M) for the TPAS plans, a difference which was significant for both. TPAS plans were preferred in 70.8% (A) and 65% (M) of cases with a significantly higher average preference of 2.1 (A) and 1.9 (M) vs 1.6 (A) and 1.2 (M) for the manual plans.
Conclusion
Plans generated by TPAS were often preferred to their historical controls, the latter of which are already highly selected for quality. These results support greater integration of TPAS into clinical planning workflows.