深度学习在预测结直肠癌淋巴结状态中的现状与挑战  

Current status and challenges of deep learning in predicting lymph node status in colorectal cancer

作　　者：赵婉婷 张广文 高慧[1] 张劲松 ZHAO Wanting;ZHANG Guangwen;GAO Hui;ZHANG Jinsong(Medical School of Yan'an University,Yan'an 716000,China;Department of Radiology,Xijing Hospital,Air Force Medical University,Xi'an 710032,China)

机构地区：[1]延安大学医学院,延安716000 [2]空军军医大学西京医院放射诊断科,西安710032

出　　处：《磁共振成像》2025年第1期222-227,共6页Chinese Journal of Magnetic Resonance Imaging

基　　金：国家自然科学基金项目(编号:82371918);西京医院临床新技术项目(编号:2024XJSY43)。

摘　　要：结直肠癌是最常见的胃肠道恶性肿瘤之一。明确结直肠癌淋巴结状态,对制订个性化治疗方案和评估预后具有重要意义。相比于专科医师视觉评估及影像组学方法,以神经网络为基础的深度学习方法由于具有自动化、高灵活性和可扩展性等优势,在结直肠癌淋巴结状态评估中具有广阔的应用前景。因此,本文将对采用深度学习方法基于CT、MRI及数字病理图片预测结直肠癌淋巴结状态的研究进展进行综述,并探讨这一领域的研究前景,以期为结直肠癌患者精准预测淋巴结状态提供新的方法和参考。Colorectal cancer(CRC)is one of the most prevalent malignant neoplasms within the gastrointestinal tract.Clarifying the status of lymph node involvement in CRC is essential for formulating personalized treatment strategies and evaluating prognosis.Compared to visual assessments by specialists and radiomics techniques,the neural network based deep learning(DL)approaches with the characteristics of high automaticity,adaptability,and scalability has demonstrated promising potential for evaluating lymph node status in CRC.Therefore,this article will provide a comprehensive review on the application of DL methods in predicting the lymph node status in CRC patients with CT,MR,and digital pathology images,and explore future research directions in this field,with the objective of providing novel methodologies and references for enhancing the accuracy of lymph node status prediction in CRC patients.

关 键 词：结直肠癌 淋巴结转移 磁共振成像 深度学习 预测 

分 类 号：R445.2[医药卫生—影像医学与核医学] R735.37[医药卫生—诊断学]