CT图像纹理分析评估胰腺神经内分泌肿瘤的病理分级

Texture analysis of CT images used to assess pathological grades of pancreatic neuroendocrine neoplasms

目的:探讨CT图像纹理分析对胰腺神经内分泌肿瘤(Pancreatic neuroendocrine neoplasm,PNEN)病理分级的诊断价值。方法:回顾性分析109例经手术或病理活检确诊为胰腺神经内分泌肿瘤患者的CT动脉期及门脉期图像,ITK-Snap软件手动勾画感兴趣区(ROI),A.K.软件提取ROI纹理特征,采用最小绝对收缩与选择算子(The least absolute shrinkage and selection operator,LASSO)降维运算获取纹理特征及其受试者工作特征曲线下面积(AUROC)。结果:筛选出动脉期纹理特征共5个,分别为熵(GLCMEntropy_AllDirection_offset7_SD)、峰度(kurtosis)、肿瘤在三维空间内的最大直径(Maximum 3D diameter)、变换后形成的正态曲线内0.025分位处的数值(Quantile0.025)及肿瘤表面积与体积的比值(Surface volume ratio),其AUROC分别为0.715、0.529、0.724、0.672及0.698,门静脉期特征2个,为肿瘤在三维空间内的最大直径及肿瘤表面积与体积的比值,其AUROC分别为0.722及0.703。结论:CT图像纹理分析可用于判断PNEN的病理分级。

Objective： To evaluate the accuracy of the texture analysis to determinate the pathological grades of pancreatic neuroendocrine neoplasms（PNEN）. Methods： 109 cases of pancreatic neuroendocrine neoplasms, confirmed by surgery or patho- logical biopsy, were retrospectively enrolled in our study. Both arterial phase and portal vein phase CT images of including patients were manual sketch the region of interest （ROI） by ITK Snap software. A.K. software was used for texture extraction, and R software with The least absolute shrinkage and selection operator （LASSO） was used for calculation. Results： In aterial phase, 5 texture features were selected, including maximum 3D diameter, kurtosis, GLCMEntropy_AllDirection_offsetT_SD, quantile0.025 and Surface volume ratio, with AUROC of 0.715, 0.529, 0.724, 0.672 and 0.698, respectively. In portal vein phase, 2 texture features were selected including maximum 3D diameter and surface volume ratio, with AUROC of 0.722 and 0.703, respectively. Conclusions： Texture analysis of CT images can be used to evaluate the pathological classification of PNEN.