CT纹理特征鉴别膀胱尿路上皮癌病理级别的效能研究  

Study on the efficacy of CT texture features in distinguishing the pathological grade of bladder urothelial carcinoma

作　　者：张雨墨 刘丽 王加良 邓珊 杨伟振 ZHANG Yu-mo(Mudanjiang Medical University,Mudanjiang 157011,China)

机构地区：[1]牡丹江医学院 [2]牡丹江医学院附属红旗医院放射线科 [3]牡丹江医学院附属红旗医院药学部,黑龙江牡丹江157011

出　　处：《牡丹江医学院学报》2023年第6期82-87,共6页Journal of Mudanjiang Medical University

基　　金：牡丹江医学院附属红旗医院研究生导师科研专项计划项目(YJSZX2022125);黑龙江省省属高等学校基本科研业务费科研项目(2019-KYYWF-0983)。

摘　　要：目的探讨CT纹理特征参数对于区分不同病理级别膀胱尿路上皮癌(urothelial carcinoma of bladder,UCB)的价值,为临床在治疗方式的选择及预后评估等提供帮助。方法本研究选取2020年3月到2022年9月在牡丹江医学院附属红旗医院经病理证实为UCB的患者60名,所有患者术前均接受CT尿路成像(CT urography,CTU)检查。将CTU图像导入MaZda软件并逐层勾画感兴趣的区域(ROI),即感兴趣的体积(VOI),行纹理分析以获得定量纹理参数。运用FPM法筛选出最佳纹理参数,将筛选出的纹理参数进行差异显著性检验,进一步绘制受试者工作特征(receiver operating characteristic curve,ROC)曲线,得出曲线下面积(area under the curve,AUC),评估纹理参数的诊断效能。将有统计学意义的不同时相的纹理参数分别构建模型及其联合模型,采用Hosmer-Lemeshow检验对模型的拟合优度进行检验,并绘制ROC曲线,得到其AUC值。结果所有提取的纹理参数经过显著性检验,皮质期、实质期、排泄期分别获得3个、4个、8个有统计学意义的纹理参数,其中,排泄期纹理参数Horzl_ShrtREmp具有最好的诊断效能,AUC值为0.679(P<0.05);在所有拟合度良好的模型中,以皮质期+实质期+排泄期模型的诊断效能最高,AUC值为0.907(P<0.05)。结论CT纹理特征参数对于区分UCB不同病理级别具有一定价值,可以为临床医生提供基于不同病理级别下疾病严重程度评价、治疗方式选择及预后评估的帮助。Objective To explore the value of CT texture parameters in distinguishing different pathological grades of urothelial carcinoma of the bladder(UCB),and then to provide help for clinical treatment and prognosis evaluation.Methods In this study,60 patients with UCB confirmed by pathology in Hongqi Hospital affiliated with Mudanjiang Medical College from March 2020 to September 2022 were selected,including 30 cases of low grade and 30 cases of high grade.All patients were examined by CTU before the operation.CTU images were imported into MaZda software,and a three-dimensional ROI,namely VOI,was drawn on MaZda software,and texture analysis was carried out to obtain quantitative texture parameters.The best texture parameters are selected by the FPM method.Using SPSS software to test the significance of the selected texture parameters,further draw the ROC curve,calculate the AUC value,evaluate the diagnostic efficiency of texture parameters for pathological grading,and calculate the best critical value.The texture parameters of different periods are statistically significant in building the model and its joint model.The goodness of fit of the model is tested by the Hosmer-Lemeshow test,and the ROC curve is drawn to get its AUC value.Results After the significance test,3,4,and 8 statistically significant texture parameters were obtained in the cortical stage,medulla stage,and excretory stage,respectively.Among them,the excretory texture parameter Horzl_ShrtREmp had the best diagnostic efficiency,and the AUC value under the ROC curve was 0.679(P<0.05);Among all the models with good fit,the diagnostic efficiency of cortical phase+medulla phase+excretory phase model was the highest,and the AUC value was 0.907(P<0.05).Conclusion CT texture feature parameters have a certain value in distinguishing different pathological grades of UCB and can help clinicians in disease severity evaluation,treatment selection,and prognosis evaluation based on different pathological grades.

关 键 词：膀胱癌 纹理分析 病理级别 

分 类 号：R445.2[医药卫生—影像医学与核医学]