目的 探讨支持向量机在CT鉴 别诊断肾脏上皮样血管平滑肌脂肪瘤 (epithelioid angiomyolipoma, EAML) 的CT与肾透明细胞癌(clear cell renal cell carcinoma, ccRCC)中的应用价值。 方法 搜集70例经病理证实的肾脏肿瘤 (EAML、ccRCC病变各35例)，采用支持向 量机法综合分析其CT特征表现，判定其 所属类型。结果 支持向量机法(support vector machine，SVM)对EAML病变的诊断 正确率为100％；对ccRCC病变的诊断正 确率为94.59％；总体平均判别正确率为 97.14％；训练集诊断正确率为97.30％； 测试集诊断正确率为96.97％；与bagging 和adaboost分类算法诊断符合率相接 近。结论 支持向量机法有助于CT鉴别诊 断EAML和ccRCC，可用于辅助日常阅片工 作，尤其是年轻医师或基层医院医师的工 作。

Objective To evaluate the application value of support vector machines in the differential diagnosis of renal epithelioid angiomyolipoma (EAML) and clear cell renal cell carcinoma (ccRCC). Methods In this study, 70 cases of renal tumors confirmed by pathology (35 cases of EAML and ccRCC) were collected, and the characteristics of CT features were analyzed by support vector machine method. Results The diagnostic accuracy of support vector machine (SVM) for EAML lesions was 100%.The diagnostic accuracy of ccRCC lesions was 94.59%. The overall average judgment accuracy was 97.14%. The diagnostic accuracy of the training set was 97.30%. The diagnostic accuracy of the test set was 96.97%. The coincidence rate is similar to bagging and adaboost classification algorithm. Conclusion Support vector machine method is helpful for CT differential diagnosis of EAML and ccRCC, which can be used to assist in daily work, especially for young doctors or primary hospital physicians.