MRI影像组学在混合型肝癌与肝内胆管细胞癌鉴别诊断中的应用

Application of MRI Radiomics in the Differential Diagnosis of Mixed Liver Cancer and Intrahepatic Cholangiocarcinoma

目的探讨MRI影像组学对混合型肝癌与肝内胆管细胞癌的鉴别诊断效能。资料与方法回顾性选取东方肝胆外科医院2019年2月—2022年2月312例原发性肝癌(术后病理为肝内胆管细胞癌224例,混合型肝癌88例)的临床信息及术前肝脏MRI影像学资料,按照2∶1分为训练集208例与验证集104例。比较两组的临床参数并建立临床模型;通过MRI影像提取影像组学特征参数并建立影像组学模型。绘制受试者工作特征曲线分析不同模型对两组的鉴别诊断效能。结果多因素分析显示性别(OR=1.663,95%CI 1.122~2.466)、乙型病毒性肝炎(OR=2.362,95%CI 1.515~3.686)、肝硬化(OR=1.308,95%CI 1.032~1.655)是鉴别肝内胆管细胞癌和混合型肝癌的影响因素(P<0.05)。基于Logit(P/1-P)的回归评估模型建立的临床、影像组学及综合模型对两组均有较好的鉴别诊断效能,曲线下面积分别为0.821、0.806、0.923,综合模型的曲线下面积高于临床模型及影像组学模型(P均<0.05)。在验证集中,3个模型的曲线下面积两两比较,差异均无统计学意义(P均>0.05)。结论MRI影像组学对肝内胆管细胞癌和混合型肝癌有较高的鉴别诊断价值,配合临床危险因素可进一步提高鉴别诊断的准确性。

Purpose To explore the differential diagnostic efficacy of MRI radiomics in mixed liver cancer and intrahepatic cholangiocarcinoma.Materials and Methods The clinical information and preoperative liver MRI imaging data of 312 patients with primary liver cancer in Eastern Hepatobiliary Surgery Hospital from February 2019 to February 2022 were retrospectively selected.The postoperative pathology showed that all patients included 224 cases with intrahepatic cholangiocarcinoma and 88 cases with mixed liver cancer.The training set(n=208 lesions)and validation set(n=104 lesions)were allocated according to the ratio of 2∶1.The clinical parameters of intrahepatic cholangiocarcinoma and mixed liver cancer were compared and clinical models were established,and the characteristic parameters of image omics were extracted from MRI radiomics and the radiomic model was established.The diagnostic efficacy of different models between intrahepatic cholangiocarcinoma and mixed liver cancer were analyzed via the receiver operating characteristic curve.Results Multivariate analysis showed that gender(OR=1.663,95%CI 1.122-2.466),hepatitis B(OR=2.362,95%CI 1.515-3.686),and cirrhosis(OR=1.308,95%CI 1.032-1.655)were the influential factors in differentiating intrahepatic cholangiocarcinoma from mixed hepatocellular carcinoma(P<0.05).Based on Logit(P/1-P)regression evaluation model,the established clinical model,radiomics model and comprehensive model had good differential diagnostic efficacy for mixed liver cancer and intrahepatic cholangiocarcinoma,and area under curve were 0.821,0.806 and 0.923,respectively.The area under curve of comprehensive model was significantly higher than that of clinical model and image omics model(all P<0.05).In the validation set,the area under curve of the three models were compared in pairs,and the differences were not statistically significant(P>0.05).Conclusion MRI radiomics has high value in the differential diagnosis of intrahepatic cholangiocarcinoma and mixed liver cancer,further improving the accuracy of differential diagnosis via combining with clinical risk factors。