CT特征联合纹理分析对肾上腺sPCC和ACC的鉴别诊断价值

Diagnostic value of CT features combined with texture analysis in differentiating sPCC from ACC

目的探讨CT特征联合纹理分析鉴别诊断肾上腺寂静性嗜铬细胞瘤(sPCC)和肾上腺皮质癌(ACC)的价值。方法分析肾上腺sPCC患者31例(A组)和ACC患者19例(B组)常规CT图像特征。在CT增强图像对病灶逐层勾画,提取CT纹理特征,选择优化的特征子集,绘制CT特征模型、CT纹理特征模型、CT特征联合纹理特征模型的ROC曲线,分析模型的诊断效能。结果两组病灶在位置、形状、有无血管影、最大径、动脉期平均CT值、动脉期峰值CT值和静脉期平均CT值的差异均有统计学意义(P<0.05)。CT特征联合纹理特征模型的AUC、一致率、灵敏度、特异度、阳性预测值和阴性预测值分别为0.988、0.940、0.935、0.947、0.967和0.900。结论 CT特征联合纹理分析对于鉴别诊断肾上腺sPCC和ACC有较高的价值。

Objective To explore the diagnostic value of CT features combined with texture analysis in differentiating adrenal silent pheochromocytoma(sPCC) from adrenocortical carcinoma(ACC).Methods Clinical and conventional CT imaging features of 31 patients with adrenal sPCC(group A) and 19 patients with ACC(group B) were analyzed.The lesions were drawn in the enhanced CT images, CT texture features were extracted, optimized feature subsets were selected, and the ROC curves of CT feature model, CT texture feature model, and CT feature combined with texture feature model were drawn to analyze the diagnostic efficacy.Results There were significant differences between the two groups in the lesion location, shape, vascular shadow, maximum diameter, mean CT attenuation of arterial phase, peak CT attenuation of arterial phase, and mean CT attenuation of venous phase(P<0.05).The AUC,accuracy, sensitivity, specificity, positive predictive value and negative predictive value of CT features combined with texture feature model were 0.988,0.940,0.935,0.947,0.967 and 0.900,respectively.Conclusion CT features combined with texture analysis has high diagnostic value in differentiating adrenal sPCC from ACC.