Progress in Medical Physics, Год журнала: 2024, Номер 35(4), С. 205 - 213

Опубликована: Дек. 31, 2024

Purpose: This study explores the potential of artificial intelligence (AI) in optimizing radiotherapy protocols for personalized cancer treatment.Specifically, it investigates role AI-based segmentation tools improving accuracy and efficiency across various anatomical regions.Methods: A dataset 500 anonymized patient computed tomography scans from Chungbuk National University Hospital was used to develop validate AI models segmenting organs-atrisk.The were tailored five regions: head neck, chest, abdomen, breast, pelvis.Performance evaluated using Dice Similarity Coefficient (DSC), Mean Surface Distance, 95th Percentile Hausdorff Distance (HD95). Results:The achieved high large, well-defined structures such as brain, lungs, liver, with DSC values exceeding 0.95 many cases.However, challenges observed smaller or complex structures, including optic chiasm rectum, instances failure infinity HD95.These findings highlight variability performance depending on complexity structure size.Conclusions: demonstrate significant streamline workflows, reduce inter-observer variability, enhance treatment accuracy.Despite integration enables dynamic, patient-specific adaptations changes, contributing more precise effective treatments.Future work should focus refining anatomically validating these methods diverse clinical settings.

Язык: Английский