基于CT影像组学预测肝细胞癌微血管侵犯准确性的Meta分析

CT Radiomics for the Preoperative Accuracy of Microvascular Invasion in Hepatocellular Carcinoma:A Meta-Analysis

目的通过Meta分析评估基于CT影像组学构建模型预测肝细胞癌(Hepatocellular Carcinoma,HCC)发生微血管侵犯(Microvascular Invasion,MVI)的准确性。方法在Web of Science、Embase、PubMed、Cochrane、中国知网、万方数据库和CBM数据库中收集相关的临床试验文献,对其进行筛选汇总分析。使用统计学软件对纳入文献进行质量以及偏倚风险评估,并且合并敏感度和特异性、阳性似然比、阴性似然比和诊断比值比以及各自的95%CI;绘制累加受试者工作特征曲线,以获得Cochran-Q指数和曲线下面积(Area Under Curve,AUC),使用敏感度分析检测潜在的异质性来源并通过Deek’s检验评估是否存在发表偏倚。结果共纳入11篇文献,汇总合并敏感度、合并特异性、合并阳性似然比、合并阴性似然比、合并诊断比值比、AUC分别为0.84(95%CI:0.79~0.89)、0.82(95%CI:0.76~0.86)、4.67(95%CI:3.57~6.11)、0.19(95%CI:0.14~0.26)、24.53(95%CI:15.40~39.07)、0.90(95%CI:0.87~0.92)。结论基于CT影像组学模型对HCC发生MVI有较高的预测效能,可以作为术前的无创评估方法之一,有较好的临床应用前景。

Objective To evaluate the accuracy of a model constructed based on CT radiomics in predicting microvascular invasion(MVI)of hepatocellular carcinoma(HCC)by using a Meta-analysis.Methods Relevant clinical trial literatures were collected from Web of Science,Embase,PubMed,Cochrane,CNKI,Wanfang and CBM databases,and screened and summarized for analysis.Statistical software was used to assess the quality and bias risk of the included literature,and sensitivity and specificity,positive likelihood ratio,negative likelihood ratio,diagnostic ODDS ratio and their respective 95%CI were combined.Summary receiver operating characteristic curve were plotted to obtain the Cochran-Q index and area under curve(AUC).Sensitivity analysis was used to detect potential sources of heterogeneity and assess the presence of publication bias through Deek’s test.Results A total of 11 literatures were included.The combined sensitivity,combined specificity,combined positive likelihood ratio,combined negative likelihood ratio,combined diagnostic ODDS ratio and AUC were 0.84(95%CI:0.79-0.89),0.82(95%CI:0.76-0.86),4.67(95%CI:3.57-6.11),0.19(95%CI:0.14-0.26),24.53(95%CI:15.40-39.07)and 0.90(0.87-0.92).Conclusion CT radiomics model has high predictive efficacy for MVI of HCC,which can be used as one of the noninvasive preoperative evaluation methods,and has a good clinical application prospect.