摘要
目的探讨基于T2WI序列腰椎MRI影像组学诊断骨质疏松症的效能及可行性。材料与方法回顾性分析2022年12月至2023年3月期间于本院行腰椎MRI检查患者共计291例,在T2WI矢状位图像上逐层勾画感兴趣区(region of interest,ROI),从1455个腰椎的MRI图像中提取放射组学特征,将样本按8∶2随机分为训练组(n=233)和测试组(n=58)。利用最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)算法降低数据维度后选择特征,利用逻辑回归(logistics regression,LR)建立预测骨质疏松的临床模型、放射组学模型及联合模型;利用受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve,AUC)、准确度、特异度、敏感度、阳性预测值和阴性预测值等指标评估组合模型的性能,使用DeLong检验比较模型间的预测效能,绘制模型的校准曲线并采用Hosmer-Lemeshow验证模型的拟合优度,采用决策曲线分析(decision curve analysis,DCA)评估各个模型的临床价值。结果临床模型、影像组学模型及联合模型在训练组中的AUC分别为0.791[95%置信区间(confidence interval,CI):0.733-0.849]、0.879(95%CI:0.833-0.925)、0.893(95%CI:0.853-0.934)。在测试组中的AUC分别为0.805(95%CI:0.676-0.935)、0.913(95%CI:0.841-0.985)、0.904(95%CI:0.825-0.984)。DeLong检测结果显示联合模型与临床模型差异具有统计学意义(P<0.05),联合模型与影像组学模型差异无统计学意义(P>0.05)。Hosmer-Lemeshow检验显示临床模型、放射组学模型及联合模型均校正良好(P=0.250,0.753,0.575)。DCA结果显示影像组学模型、联合模型预测骨质疏松的临床价值均优于临床模型。结论基于腰椎T2WI构建的影像组学模型具有客观、准确诊断骨质疏松症的潜力。
Objective:To investigate efficacy of radiomics on the lumbar spine MRI based on T2WI sequences in identifying osteoporosis.Materials and Methods:A retrospective analysis was conducted on a total of 291 patients who underwent lumbar spine MRI examinations at our hospital between December 2022 and March 2023.Regions of interest(ROI)were delineated layer by layer on the sagittal T2WI images.Radiomic features were extracted from the MR images of 1455 lumbar vertebrae.The samples were randomly divided into a training group(n=233)and a test group(n=58)at an 8∶2 ratio.The least absolute shrinkage and selection operator(LASSO)was used to reduce data dimensionality and select features.Logistic regression(LR)was employed to establish clinical models,radiomic models,and a combined model for predicting osteoporosis.The performance of the composite models was evaluated using metrics such as the area under the curve(AUC)of receiver operating characteristic(ROC),accuracy,specificity,sensitivity,positive predictive value,and negative predictive value.DeLong test was used to compare the predictive performance among the models.Calibration curves for the models were plotted,and Hosmer-Lemeshow test was applied to assess model fit.Decision curve analysis(DCA)was used to evaluate the clinical utility of each model.Results:In the training group,the AUCs for the clinical model,radiomic model,and combined model were 0.791[95%confidence interval(CI):0.733-0.849],0.879(95%CI:0.833-0.925),and 0.893(95%CI:0.853-0.934),respectively.In the test group,the AUCs were 0.805(95%CI:0.676-0.935),0.913(95%CI:0.841-0.985),and 0.904(95%CI:0.825-0.984),respectively.DeLong test results indicated that there was a statistically significant difference between the combined model and the clinical model(P<0.05),while there was no statistically significant difference between the combined model and the radiomic model(P>0.05).The Hosmer-Lemeshow test showed that the models were well calibrated(P=0.250,0.753,0.575).The results of DCA demonstrated that both the r
作者
康嗣如
田荣华
KANG Siru;TIAN Ronghua(Department of Radiology,Xiaogan Hospital Affiliated to Wuhan University of Science and Technology,Xiaogan 432000,China)
出处
《磁共振成像》
CAS
CSCD
北大核心
2023年第11期121-127,共7页
Chinese Journal of Magnetic Resonance Imaging
基金
孝感市自然科学计划项目(编号:XGKJ2022010002)。
关键词
腰椎
骨质疏松
磁共振成像
影像组学
lumbar spine
osteoporosis
magnetic resonance imaging
radiomics
