摘要
目的探讨基于X线腹部平片(KUB)、低剂量CT影像组学列线图预测攀枝花泌尿系统结石成分价值,以期为临床明确结石成分、制定治疗方案提供依据。方法选取2019年6月至2022年6月我院96例泌尿系统结石患者作为研究对象,均行KUB、低剂量CT检查,根据术后结石成分红外光谱检测法定性结果分为尿酸结石组(n=83)和非尿酸结石组(n=13),采用A.K.软件提取CT影像组学特征并构建影像组学标签,Logistic回归方程筛选泌尿系统结石成分预测因子,构建列线图模型,ROC曲线评估模型预测效能,行内外部验证。结果尿酸结石组和非尿酸结石组BMI、血UA、尿pH值、血Scr、高血压、糖尿病比较存在显著差异(P<0.05);血UA、尿pH值、高血压、糖尿病、影像组学标签评分是泌尿系统结石成分影响因素(P<0.05);ROC曲线显示,列线图模型在训练集和验证集人群中AUC分别为0.915、0.915,该模型预测在训练集和验证集人群中结果与实际观察结果之间有很好相关性,DCA曲线显示在范围0.8~0.9、0.3~0.9内,该模型在训练集和验证集人群中净获益值较好。结论基于KUB、低剂量CT影像组学列线图可用于攀枝花泌尿系统结石成分预测评估中,临床可通过模型相关因素早期预测结石成分,以针对性制定治疗方案。
Objective To investigate the value of predicting urinary calculi composition in Panzhihua based on plain abdominal radiographs(KUB)and low-dose CT image,so as to provide a basis for clinical determination of the composition of stones and formulation of treatment plans.Methods 96 patients with urinary system stones in our hospital from June 2019 to June 2022 were selected as research objects.All patients underwent KUB and low-dose CT examination.According to the qualitative results of postoperative modified Mauer method,they were divided into uric acid calculi group(n=83)and non-uric acid calculi group(n=13).The A.K.software was used to extract the CT imageomics characteristics and construct the imageomics tags,the Logistic regression equation was used to screen the predictors of urinary system stone composition,and the nomogram model was constructed.The ROC curve was used to evaluate the prediction efficiency of the model,and the internal and external validation was performed.Results There were significant differences in BMI,blood UA,urine pH,blood Scr,hypertension and diabetes between uric acid stone group and non-uric acid stone group(P<0.05).Blood UA,urine pH,hypertension,diabetes,and imaging label score were the factors affecting urinary calculus composition(P<0.05).ROC curve showed that the AUC of the line graph model in the training set and the verification set population was 0.915 and 0.915,respectively.There was a good correlation between the results predicted by the model in the training set and the verification set population and the actual observation results,and the DCA curve was displayed in the range of 0.8~0.9 and 0.3~0.9.The net benefit value of this model is good in both training set and verification set.Conclusion KUB and low-dose CT image histogram can be used to predict and evaluate urinary stone composition in Panzhihua.In clinic,early prediction of stone composition can be made based on factors related to the model,so as to make targeted treatment plan.
作者
刘范林
蒋群
杨小君
马清明
LIU Fan-lin;JIANG Qun;YANG Xiao-jun;MA Qing-ming(Medical Imaging Department,Panzhihua Second People's Hospital,Panzhihua 617000,Sichuan Province,China;Department of Ultrasound,Panzhihua Second People's Hospital,Panzhihua 617000,Sichuan Province,China)
出处
《中国CT和MRI杂志》
2023年第9期133-136,共4页
Chinese Journal of CT and MRI
基金
攀枝花市指导性科技计划项目(2020ZD-S-11)
四川省医学科研课题计划项目(Q18044)
关键词
X线腹部平片
低剂量CT
影像组学
列线图
泌尿系统结石
预测价值
Plain Abdominal Radiographs
Low-dose CT
Imaging Omics
A Column Diagram
Urinary Calculi
Predictive Value
