Abstract: Tongue squamous cell carcinoma is a subtype of oral squamous cell carcinoma with a high incidence. Because this region involves complex anatomical structure and rich lymphatic reflux，it has a high rate of cervical lymph node metastasis and a poor prognosis. As a solid tumor，conventional imaging methods often can not reflect its internal heterogeneity clearly. Radiomics can reflect the intra-tumor heterogeneity by extracting advanced quantitative imaging features in a high-through manner，and can be used for further analysis by integrating machine learning，which is used to develop image feature classification or prediction models related to tumor phenotype or gene-protein features，aiding clinical decision making. In this paper，the research progress on the application of machine learning-based magnetic resonance imaging（MRI）radiomics was reviewed，concerning the prediction of the degree of pathological differentiation，cervical lymph node metastasis，effect and prognosis of tongue squamous cell carcinoma in recent years，in order to provide refence for precision medicine of patients with tongue squamous cell carcinoma.

Key words: oral tongue squamous cell carcinoma, magnetic resonance imaging, radiomics, machine learning

摘要： 舌鳞状细胞癌（以下简称“舌癌”）是口腔鳞状细胞癌中发病率较高的亚型。由于此区域解剖结构复杂、淋巴回流丰富，因而舌癌具有颈淋巴结转移率高、预后差的特点。舌癌作为一种实体性肿瘤，常规的医学影像图像往往不能较好反映肿瘤内部异质性。影像组学可通过一种高通量方式提取高级定量图像特征，捕获肿瘤内部异质性，并结合机器学习算法进一步分析，用来建立与肿瘤表型或基因-蛋白质特点相关的图像特征分类或预测模型，辅助临床决策制定。文章就近年来基于机器学习的核磁共振成像（MRI）影像组学在预测舌癌病理分化程度、颈淋巴结转移、疗效及预后等方面应用的研究进展做一阐述，以期为舌癌患者精准医疗提供参考。

关键词: 舌鳞状细胞癌, 核磁共振成像, 影像组学, 机器学习