Dosimetric Comparison of Organs At Risk Between Artificial Intelligence-Based Auto-Contouring and Manual Contouring for High-Risk Prostate Cancer Radiotherapy: A Retrospective Study.
Background and aim This study compared the dose volume histogram (DVH) parameters of the organs at risk (OARs) between an artificial intelligence (AI)-based auto-contouring method and conventional man
- p-value p=0.067
- 95% CI -0.87 to 4.47
APA
Kawaguchi M, Sugahara S, et al. (2025). Dosimetric Comparison of Organs At Risk Between Artificial Intelligence-Based Auto-Contouring and Manual Contouring for High-Risk Prostate Cancer Radiotherapy: A Retrospective Study.. Cureus, 17(9), e91810. https://doi.org/10.7759/cureus.91810
MLA
Kawaguchi M, et al.. "Dosimetric Comparison of Organs At Risk Between Artificial Intelligence-Based Auto-Contouring and Manual Contouring for High-Risk Prostate Cancer Radiotherapy: A Retrospective Study.." Cureus, vol. 17, no. 9, 2025, pp. e91810.
PMID
41080339
Abstract
Background and aim This study compared the dose volume histogram (DVH) parameters of the organs at risk (OARs) between an artificial intelligence (AI)-based auto-contouring method and conventional manually drawn contours in radiation treatment planning for high-risk prostate cancer. Methods Fifteen treatment plans for high-risk prostate cancer treated with 76 Gy in 38 fractions using volumetric modulated arc therapy (VMAT) were retrospectively analyzed. For each patient, both manual contouring (MC) by an experienced radiation oncologist and automated contouring (AC) by OncoStudio (Oncosoft, Seoul, Republic of Korea) were performed, and then the DVH parameters of the OARs were compared between MC and AC. Primary outcome measures were dose constraints (V20 to V75) for the rectum, bladder, and the femoral heads. T-test statistical analysis was performed. Results The analysis showed no statistically significant differences between the MC and AC groups in all the DVH parameters for OARs (p>0.05). The mean difference in rectum V50 was 1.8% (95%CI: -0.87 to 4.47%, p=0.178), and the mean difference in bladder V70 was 1.6% (95%CI: -0.12 to 3.32%, p=0.067). The effect sizes R² ranged from 0.001 to 0.115, and the observed differences were small and clinically acceptable. Conclusion In high-risk prostate cancer radiotherapy planning, AC with OncoStudio demonstrated non-inferiority in the OARs dose assessment compared to MC. There were no statistically significant differences in the DVH parameters of all the OARs, and the observed differences were clinically acceptable. The introduction of AC is expected to streamline the OARs contouring process while maintaining clinical equivalence.