目的 探讨CT图像纹理分析在预测胸腺上皮性肿瘤(TETs)简化病理分型中的价值.方法 回顾性分析经手术或病理证实的114例TETs患者的CT图像,将TETs分型简化为低危胸腺瘤(LRT)、高危胸腺瘤(HRT)和胸腺癌(TC);提取肿瘤CT平扫及增强扫描各期的纹理参数,对纹理参数进行筛选、降维后求得加权影像组学评分(Rad-score)值,使用受试者工作特征(ROC)曲线分析纹理特征的预测效能.结果 114例TETs患者中,LRT45例,HRT44例,TC25例.基于CT图像纹理分析,在鉴别LRT与HRT和TC时,CT平扫、动脉期和静脉期的曲线下面积(AUC)分别为0.776、0.885和0.761.在鉴别HRT与TC时,CT平扫、动脉期和静脉期的AUC分别为0.828、0.808和0.804.在鉴别胸腺瘤与TC时,CT平扫、动脉期和静脉期的AUC分别为0.808、0.769和0.774.结论 CT图像纹理分析可以作为预测TETs简化病理分型的有效辅助手段,有助于为TETs患者制订个性化的治疗方案,其中CT增强扫描动脉期纹理参数具有最高的鉴别诊断效能.
Objective To investigate the value of CT imaging texture analysis in predicting simplified pathological types of thymic epithelial tumors(TETs).Methods The CT data from 114 patients with TETs confirmed by surgical or pathology were analyzed retrospectivel,and the types of TETs were divided into three groups,including low-risk thymoma(LRT)group,high-risk thymoma(HRT)group,and thymic carcinoma(TC)group.First,the texture parameters of CT images were extracted,and then the weighted Rad-score values were obtained,and the predictive performance of the texture features was evaluated by using the receiver operating characteristic(ROC)curve.Results There were 114 TETs patients,including 45 patients with LRT,44 patients with HRT,and 25 patients with TC.Based on CT imaging texture analysis,the area under the curve(AUC)in differentiating LRT and HRT or TC via CT plain scan,arterial phase,and venous phase were 0.776,0.885,and 0.761,respectively.In differentiating HRT from TC,the AUC of CT plain scan,arterial phase,and venous phase were 0.828,0.808,and 0.804,respectively.In differentiating thymoma from TC,the AUC of CT plain scan,arterial phase,and venous phase were 0.808,0.769,and 0.774,respectively.Conclusion CT imaging texture analysis can serve as an effective auxiliary tool for predicting the simplified pathological types of TETs,helping to develop personal-ized treatment plans for TETs patients.CT enhanced scanning of arterial phase texture parameters has the highest differential diag-nostic efficiency.