基于胸部CT扫描数据的骨质疏松诊断方法初步研究

A preliminary study on the diagnostic method of osteoporosis based on chest CT scan data

目的探讨使用机会性胸部CT扫描数据中胸椎T1-T12的CT均值进行二元逻辑回归分析以诊断骨质疏松的可行性。方法本文回顾性研究收集了326名接受胸部CT的体检者(平均年龄64岁±10岁;男性111名,女性218名)胸椎T1-T12的CT均值,以双能X线骨质疏松的诊断结果为金标准。分别将T1-T12的CT均值,T1-T12的CT均值结合性别、年龄、身高、体质量和BMI作为输入建立二元逻辑回归模型。随后将第二组数据按照性别、年龄和BMI进行分组建模,讨论各组模型对骨质疏松的预测效能。结果以T1-T12的CT均值作为二元逻辑回归模型的输入,模型预测准确性为0.763(249/326);以T1-T12的CT均值结合性别、年龄、身高、体质量和BMI作为二元逻辑回归模型的输入,模型预测准确性为0.825(269/326)。结论利用胸部CT T1-T12的CT均值结合性别、年龄、身高、体质量和BMI作为辅助因素进行骨质疏松症自动诊断的方法具有较高的精度,该模型对女性(准确性为0.838),对50~59岁年龄段人群(准确性为0.826)和超重人群(准确性为0.877)有较好的诊断效能。

Objective To explore the feasibility of using the mean CT values of T1-T12 thoracic vertebra from opportunistic chest CT scan data in the diagnosis of osteoporosis based on binary logistic analysis.Methods A retrospective study was conducted to collect the mean value of thoracic T1-T12 in 326 patients(mean age 64±10;including 111 males and 218 females)who underwent chest CT.The diagnostic results of dual energy X-ray osteoporosis were used as the gold standard.The mean CT values of T1-T12 and the mean CT values of T1-T12 combined with gender,age,height,weight and BMI were as inputs to establish binary logistic models.The second group of data were grouped according to gender,age and BMI,and the predictive efficacy of each model for osteoporosis was discussed.Results The model prediction accuracy was 0.763(249/326)with the mean CT value of T1-T12 as the input of the binary logistic model.The model prediction accuracy was 0.825(269/326)with CT mean values of T1-T12 combined with gender,age,height,weight and BMI as input of binary logistic model.Conclusions The method of using the CT mean of chest CT T1-T12 combined with gender,age,height,weight and BMI as cofactors for automatic diagnosis of osteoporosis has high accuracy.The model shows good diagnostic efficacy for women(0.838),50-59 age group(0.826)and overweight group(0.877).