Objective: To explore the value of CT texture analysis (CTTA) in differentiating the pathological grade of urothelial carcinoma of the bladder (UCB). Methods: A total of 53 lesions from 43 patients with bladder cancer confirmed by postoperative pathology were retrospectively analyzed, including 27 cases of high-grade urothelial carcinoma (HGUC) and 26 cases of low-grade urothelial carcinoma (LGUC). All the patients took pelvic CT and enhanced scanning in the same CT scanner with same scanning parameters. Lesions on both plain and enhanced CT images were delineated on software by two radiologists to extract the corresponding volumes of interest (VOI) and then 92 parameters based on feature classes were generated. The average values of two radiologists were obtained. The difference parameters between HGUC group and LGUC group were screened by nonparametric test, and the receiver operating characteristic (ROC) was drawn. The corresponding optimal thresholds were determined and diagnostic effect was assessed. Results: Nine difference texture parameters between HGUC group and LGUC group were selected, including 5 parameters on unenhanced images, namely, skewness, root mean squared, cluster shade, zone percentage and large area high gray level emphasis. There were 4 parameters on enhanced images, namely, skewness, kurtosis, cluster shade and zone percentage. The largest area under curve of 0.840±0.058 (95% CI 0.726-0.955) was obtained from skewness generated by VOI of unenhanced images. The cut-off value of skewness was 0.186 5, which permitted the diagnosis of HGUC with sensitivity of 92.59%, specificity of 73.08%, positive predictive value of 78.13%, negative predictive value of 90.48% and accuracy of 83.02%. Conclusion: CTTA can effectively distinguish between LGUC and HGUC. Skewness from unenhanced CT images had the optimal diagnostic performance.
目的：探讨CT图像纹理分析方法在鉴别膀胱尿路上皮癌不同病理学级别中的价值。 方法：回顾性分析43例经术后病理证实的膀胱癌患者的53个病灶，其中高级别尿路上皮癌(HGUC)27个，低级别尿路上皮癌(LGUC)26个。所有患者在同一台CT机上，以同样的扫描参数进行盆腔CT平扫和增强扫描。2名影像科医师分别利用软件工具在CT平扫和增强图像上对病灶进行勾画，并获得感兴趣容积(VOI)，生成基于特征类的92个参数，取2名影像科医师测量数据的平均值。采用非参数检验筛选HGUC组和LGUC组间的差异参数，绘制其受试者工作特征(ROC)曲线，确定差异参数的最佳界值，并对诊断效果进行评价。 结果：筛选出HGUC组和LGUC组间的差异纹理参数9个，其中平扫图像参数5个，分别是偏度、均方根、集群阴暗度、区域百分比和大面积高灰度增强；增强图像参数4个，分别是偏度、峰度、集群阴暗度和区域百分比。根据平扫图像VOI获得偏度的曲线下面积最大，为0.840±0.058(95%可信区间：0.726～0.955)。偏度的最佳界值为0.186 5，诊断HGUC的敏感度为92.59%，特异度为73.08%，阳性预测值为78.13%，阴性预测值为90.48%，准确性为83.02%。 结论：基于CT图像的纹理分析方法可有效区分膀胱LGUC与HGUC，其中偏度这一参数诊断效能最佳。.
Zhonghua zhong liu za zhi [Chinese journal of oncology]. 2018 May 23 [Epub]
Z H Liu, J Y Shi, H Y Wang, H Y Ye, Z B Wang, T Yang, X Ma, X Bai
Department of Radiology, Affiliated Hospital of Changzhi Institute of Traditional Chinese Medicine, Changzhi 046000, China., Department of Radiology, Shaanxi Sengong Hospital, Xi'an 710300, China., Department of Radiology, Chinese PLA General Hospital, Beijing 100853, China., Department of Pathology, Chinese PLA General Hospital, Beijing 100853, China., Department of Urology, Chinese PLA General Hospital, Beijing 100853, China.