Application Progress of Artificial Intelligence in Radiotherapy for Rectal Cancer
10.3971/j.issn.1000-8578.2026.26.0086
- VernacularTitle:人工智能在直肠癌放射治疗中的应用进展
- Author:
Weijin ZHANG
1
;
Wenming YAN
2
Author Information
1. First Clinical Medical College, Inner Mongolia Medical University, Hohhot 010050, China.
2. Department of Radiotherapy, Affiliated Hospital of Inner Mongolia Medical University, Hohhot 010050, China.
- Publication Type:REVIEWS
- Keywords:
Artificial intelligence;
Deep learning;
Rectal neoplasms;
Radiotherapy;
Automatic target delineation
- From:
Cancer Research on Prevention and Treatment
2026;53(8):633-641
- CountryChina
- Language:Chinese
-
Abstract:
Radiotherapy is integral to theneoadjuvant and definitive treatment of rectal cancer, yet its clinical application faces challenges, such as inefficient target volume delineation, substantial interindividual heterogeneity in treatment response, and difficulties in toxicity prediction. In recent years, artificial intelligence (AI), particularly deep learning, has emerged as a promising approach to address the above challenges, offering the potential to enhance the precision and efficiency of radiotherapy considerably. Its core applications primarily include the following aspects: First, it enables the high-precision automatic segmentation of organs at risk, achieving a Dice similarity coefficient of over 0.85. Second, it facilitates intelligent plan optimization, improving time efficiency by 40%–60%. Third, it supports the construction of multimodal dose-toxicity prediction models, with an area under the curve ranging from 0.82 to 0.93.This review systematically discusses the application of AI across multiple stages of the entire radiotherapy course for rectal cancer, including imaging diagnosis, target delineation, plan optimization, toxicity prediction, and efficacy assessment. Relevant technological advancements have substantially contributed to the progression of radiotherapy toward increased precision and efficiency. Furthermore, this reviewprovides an in-depth analysis of current translational bottlenecks, such as insufficient model interpretability, data heterogeneity, and lack of multicenter validation. It aims to offer theoretical support and practical references for AI-driven precision radiotherapy in rectal cancer.
