直肠癌淋巴结转移MRI人工智能研究进展

MRI-Based Artificial Intelligence in Lymph Node Metastasis of Rectal Cancer

直肠癌是常见的消化系统恶性肿瘤,转移方式以淋巴转移为主,淋巴结转移决定治疗方案及疾病预后。目前评价直肠癌淋巴结转移主要依赖高分辨MRI,但基于MRI形态学诊断淋巴结转移的主观性强,诊断结果无法完全一致。人工智能(artificial intelligence,AI)能够深度挖掘医学图像的定量信息,为评估直肠癌淋巴结转移提供了新的途径。本文总结了近年来基于MRI的AI评估直肠癌治疗前和新辅助治疗后淋巴结转移的研究进展,并进行小结和展望,以期帮助读者了解基于直肠癌MRI评估淋巴结转移的AI研究存在回顾性研究多、数据集规模不一、模型效能参差的局限,为未来设计前瞻性、多中心、大数据的AI研究提供参考。

Rectal cancer is one of the most common malignancies in the digestive tract.Cancer cells usually disseminate from rectal tumors to distant sites via lymphatic vessels.Thus,lymph node involvement,which influences treatment and prognosis,plays a crucial role in patients with rectal cancer.High resolution MRI has been used to estimate lymph node metastasis in rectal cancer.However,the morphological criteria were influenced by the subjective judgement of different observers.Artificial intelligence(AI)can mine and learn quantitative features from medical images,thus providing a new method for us to distinguish metastatic lymph nodes.In this review,we summarize the research progress of MRI-based AI in the evaluation of nodal metastasis with rectal cancer before and after the neoadjuvant chemoradiotherapy.We further discuss the challenges and provide prospects of AI research to help researchers understand the limitations of MRI-based AI in evaluation of nodal involvement in rectal cancer and offer guidance for future prospective,multi-center,big-data AI research.