目前最常采用影像学TRG分级方法是2011年MERUCRY协作组根据pTRG提出的，但是由于缺少客观、定量的评价参数导致准确率不高，一致性较低。医学影像智能分析为直肠癌新辅助放化疗后的肿瘤退缩分级诊断提供了新的工具，采用传统的机器学习方法和近几年发展迅速的深度学习方法，将有可能实现新辅助放化疗肿瘤退缩分级状况的智能诊断，切实提高直肠癌的临床诊疗水平。.本项目拟开展前瞻性、多中心研究，应用深度学习方法建立直肠癌新辅助放化疗后肿瘤退缩分级的影像、病理、基因融合的多模态智能评估模型，实现术前对病理TRG的准确预测，以及对患者远期预后的风险评估。研究将开发一套智能化、定量化、可视化的TRG辅助诊断系统，切实提高临床医生工作效率，促进研究成果的临床实践应用转化。

At present, the most commonly used radiological TRG classification method is proposed by MERUCRY collaboration group in 2011 based on pTRG. However, due to the lack of objective and quantitative evaluation parameters, the accuracy is not high and the inter-observer consistency is low. .Intelligent analysis of medical imaging serves a new tool for tumor regression grade of rectal cancer after neoadjuvant chemoradiotherapy. It is possible to achieve intelligent diagnosis of TRG after neoadjuvant chemoradiotherapy by using traditional machine learning methods and in-depth learning methods which have developed rapidly in recent years, and to improve the level of clinical diagnosis and treatment of rectal cancer. .This project proposes to carry out a prospective, multi-center study, using in-depth learning method to establish a multi-modal intelligent evaluation model by fusion of for image, pathology and gene information TRG evaluation of rectal cancer. The multi-modal intelligent evaluation will achieve accurate preoperative prediction of pathological TRG and risk assessment of long-term prognosis of patients. The research will develop a set of intelligent, quantitative and visualized assistant diagnosis system for TRG to improve the work efficiency of clinicians, and to actively promote the transformation of clinical application of research results.