Comprehensive geometry-based plan quality score for lung VMAT

Volumetric-modulated arc therapy (VMAT) treatment planning for lung tumors requires balancing adequate planning target volume (PTV) coverage with optimal organ-at-risk (OAR) sparing. However, plan quality varies due to differences in planner expertise, preferences, and especially patient anatomy. This study introduces a Comprehensive Plan Score (CPS) for lung VMAT plans, integrating geometry-based OAR quality indices derived from overlap volume histograms (OVH) and overlap-z-histograms (OZH). CPS was developed by combining normalized target and OAR scores, including geometry-based metrics derived from OVH and OZH. Relative weights were learned from blinded pairwise plan comparisons performed by four senior radiation oncologists (ROs). A total of 35 manual versus automated lung VMAT plan pairs were evaluated, with 20 cases used for training and 15 for independent testing. Across all comparisons, ROs expressed a clear preference in 81% of cases, while 19% were judged equivalent. Inter-observer agreement was fair (Fleiss' κ = 0.36), with pairwise Cohen's κ ranging from 0.13 to 0.71. In the independent test cohort, CPS using the preference-trained weighting set matched the ROs consensus in 11/15 cases. Applying an equivalence threshold (ΔCPS = 0.01) increased agreement to 13/15 cases (86%). In conclusion, the CPS provides a more objective and geometry-based measure of lung VMAT plan quality. By quantifying achievable OAR sparing based on prior data, it supports consistent and standardized plan evaluation, potentially improving the reproducibility and overall quality of radiotherapy planning.