Clinical Validation and Quantitative Evaluation of a Novel Artificial-Intelligence Automatic Segmentation Tool in Radiotherapy Planning

1. Abstract
1.1. Purpose: Organ-at-risk (OAR) contouring for radiation treatment planning requires significant physician effort. Manual segmentation (MS) of patient organs remains a large time sink for physicians in radiation oncology. Auto-contouring systems aim to reduce this effort, but quality remains inconsistent. Artificial-intelligence auto-segmentation (AI-AS) has emerged as an alternative to atlas-based methods, promising improved results without the effort necessary to create a functional atlas. This study evaluates a novel, commercial AI-AS system for head and neck (H&N), central nervous system (CNS), and breast OAR contouring via comparison to manually delineated contours by an experienced radiation oncologist.