Novel dosimetric validation of a commercial CT scanner based deep learning automated contour solution for prostate...

OAR delineation accuracy influences: (i) a patient's optimised dose distribution (PD), (ii) the reported doses (RD) presented at approval, which represent plan quality. This study utilised a novel dosimetric validation methodology, comprehensively evaluating a new CT-scanner-based AI contouring solution in terms of PD and RD within an automated planning workflow.

20 prostate patients were selected to evaluate AI contouring for rectum, bladder, and proximal femurs. Five planning 'pipelines' were considered; three using AI contours with differing levels of manual editing (nominally none (AI_Std), minor editing in specific regions (AI_MinEd), and fully corrected (AI_FullEd)). Remaining pipelines were manual delineations from two observers (MD_Ob1, MD_Ob2). Automated radiotherapy plans were generated for each pipeline. Geometric and dosimetric agreement of contour sets AI_Std, AI_MinEd, AI_FullEd and MD_Ob2 were evaluated against the reference set MD_Ob1. Non-inferiority of AI pipelines was assessed, hypothesising that compared to MD_Ob1, absolute deviations in metrics for AI contouring were no greater than that from MD_Ob2.

Compared to MD_Ob1, organ delineation time was reduced by 24.9 min (96 %), 21.4 min (79 %) and 12.2 min (45 %) for AI_Std, AI_MinEd and AI_FullEd respectively. All pipelines exhibited generally good dosimetric agreement with MD_Ob1. For RD, median deviations were within ± 1.8 cm³, ± 1.7 % and ± 0.6 Gy for absolute volume, relative volume and mean dose metrics respectively. For PD, respective values were within ± 0.4 cm³, ± 0.5 % and ± 0.2 Gy. Statistically (p < 0.05), AI_MinEd and AI_FullEd were dosimetrically non-inferior to MD_Ob2.

This novel dosimetric validation demonstrated that following targeted minor editing (AI_MinEd), AI contours were dosimetrically non-inferior to manual delineations, reducing delineation time by 79 %.