周金发,唐继来,朱建国,李海歌.基于增强CT直方图参数术前预测高、低分化胰腺腺癌[J].中国医学影像技术,2022,38(2):215~219
基于增强CT直方图参数术前预测高、低分化胰腺腺癌
Histogram parameters based on enhancement CT for preoperative predicting pancreatic adenocarcinomas with high and low differentiation
投稿时间:2021-03-11  修订日期:2021-09-17
DOI:10.13929/j.issn.1003-3289.2022.02.013
中文关键词:  胰腺肿瘤  细胞分化  体层摄影术,X线计算机  直方图
英文关键词:pancreatic neoplasms  cell differentiation  tomography, X-ray computed  histogram
基金项目:南京医科大学科技发展基金(NMUB2020058)。
作者单位E-mail
周金发 南京医科大学第二附属医院放射科, 江苏 南京 210011  
唐继来 南京医科大学第二附属医院放射科, 江苏 南京 210011 tangjilai@163.com 
朱建国 南京医科大学第二附属医院放射科, 江苏 南京 210011  
李海歌 南京医科大学第二附属医院放射科, 江苏 南京 210011  
摘要点击次数: 1096
全文下载次数: 656
中文摘要:
      目的 评估基于增强CT直方图(CTH)定量参数术前预测高、低分化胰腺腺癌的价值。方法 回顾性观察76例经手术病理证实胰腺腺癌患者,包括38例高分化腺癌(高分化组)及38例低分化腺癌(低分化组)。对腹部增强CT动脉期及门静脉期图像进行直方图分析,获得病灶及胰腺组织的CTH定量参数,计算CTH参数比值(瘤区CTH参数/正常胰腺组织CTH参数);比较组间CTH参数比值的差异,针对差异有统计学意义的参数,采用二元logistic回归分析筛选高、低分化胰腺腺癌的独立预测因子,并分别绘制受试者工作特征(ROC)曲线,计算相应曲线下面积(AUC),评估其单独及联合预测高、低分化胰腺腺癌的效能。结果 组间动脉期最小强度、均匀度及静脉期最小强度、平均值、均方根值及均匀度比值差异均有统计学意义(P均<0.05)。二元logistic回归分析显示,动脉期最小强度、均匀度及静脉期最小强度比值为胰腺腺癌分化程度的独立预测因子(P均<0.05)。ROC曲线显示,以动脉期最小强度、动脉期均匀度及静脉期最小强度比值预测高、低分化胰腺腺癌的AUC分别为0.758、0.752及0.815,三者联合AUC为0.868。结论 基于增强CT的CTH定量参数有助于术前预测高、低分化胰腺腺癌。
英文摘要:
      Objective To explore the value of CT histogram (CTH) quantitative parameters based on enhancement CT for preoperative predicting pancreatic adenocarcinomas with high and low differentiation. Methods Preoperative abdominal enhancement CT of 76 patients with pancreatic adenocarcinoma confirmed by postoperative pathology were retrospectively analyzed, including 38 cases with high differentiation adenocarcinoma (high differentiation group) and 38 cases with low differentiation adenocarcinoma (low differentiation group). Histogram analysis was performed on abdominal arterial and portal phase CT images to obtain CTH quantitative parameters of tumors and pancreatic tissue, and the ratios of CTH parameters were calculated according to formula of CTH parameter of tumor area/CTH parameter of pancreatic tissue, and were compared between groups. Binary logistic regression analysis was performed on parameters being significant difference between groups to screen the independent predictors of pancreatic adenocarcinomas with high and low differentiation. The receiver operating characteristic (ROC) curves were drawn, the corresponding areas under the curves (AUC) were calculated to evaluate the predicting value of the above parameters for pancreatic adenocarcinomas with high and low differentiation alone and in combination, respectively. Results There were significant differences of the ratios of the minimum intensity, the uniformity in arterial phase, as well as of the minimum intensity, the mean value, root mean square value and the uniformity in portal phase between groups. Binary logistic regression showed that the ratio of the minimum intensity and the uniformity in arterial phase, also the minimum intensity in portal phase were independent predictors of pancreatic adenocarcinomas with high and low differentiation (all P<0.05). ROC curves showed that AUC of the ratio of the minimum intensity and the uniformity in arterial phase, and the minimum intensity in portal phase for predicting pancreatic adenocarcinomas with high and low differentiation was 0.758, 0.752 and 0.815, respectively, while of the combination of them was 0.868. Conclusion CTH quantitative parameters based on enhancement CT were helpful for preoperative predicting pancreatic adenocarcinomas with high and low differentiation.
查看全文  查看/发表评论  下载PDF阅读器