CT影像组学对肾上腺非结节样增生患者高血压发生风险的预测价值OA
Predictive value of computed tomography radiomics for thie risk of hypertension in patients with non-nodular adrenal hyperplasia
目的 通过计算机断层扫描(CT)影像组学探索肾上腺非结节样增生与高血压风险的关系.方法 回顾性分析452例肾上腺非结节样增生患者的CT影像数据及临床资料.利用LASSO回归模型选择影像组学特征,用单因素回归分析筛选CT形态学特征,用多因素回归分析构建影像组学模型及CT形态学模型.整合CT形态学和影像组学特征,构建综合模型并进行亚组分析.采用受试者操作特征(ROC)曲线、Delong检验、校准曲线和决策曲线评估模型性能.结果 按8∶2比例将数据分为训练集(360例)和验证集(92例),从2 446个影像组学特征中筛选出11个用于构建影像组学模型.单因素回归分析确定与高血压相关的危险因素为两侧肾上腺的面积、体积和结合部厚度.综合模型、影像组学模型和CT形态学模型在训练集中的ROC曲线下面积(AUC)分别为0.839、0.831和0.620,在验证集中的AUC分别为0.804、0.802和0.693.校准曲线、决策曲线和Delong检验显示,影像组学模型性能优于CT形态学模型(P<0.05),与综合模型无显著差异(P>0.05).亚组分析显示,模型在年龄和性别亚组中预测性能较为一致(P>0.05).结论 影像组学模型对高血压风险具有较好的预测性能,CT诊断为肾上腺非结节样增生时,其肾上腺形态学增大和纹理异质性增加可加剧高血压风险.
Objective To explore the relationship between non-nodular adrenal hyperplasia and the risk of hypertension using a com-puted tomography(CT)radiomics assessment.Methods The data of 452 patients with non-nodular adrenal hyperplasia were retrospec-tively analyzed.The Least Absolute Shrinkage and Selection Operator model was used to select the radiomics features,univariate regres-sion analysis was used to screen the CT morphological features,and multivariable regression analysis was used to construct corresponding models.By integrating CT morphological and radiomics features,comprehensive models were constructed,and a subgroup analysis was performed.The receiver operating characteristic(ROC)curve,Delong test,calibration curve,and decision curve were used to analyze and evaluate the performance of the models.Results Based on a ratio of 8∶2,360 patients were assigned to the training set and 92 were assigned to the validation set.Eleven of the 2 446 radiomics features were selected to construct the radiomics model.Six risk factors associated with hypertension were identified using univariate regression analysis:area,volume,and junction thickness of both adrenal glands.The areas under the ROC curve of the comprehensive,radiomics,and CT morphological models were 0.839,0.831,and 0.620,respectively,in the training set,and 0.804,0.802,and 0.693,respectively,in the validation set.The calibration curve,decision curve,and Delong test showed that the performance of the radiomics model was superior to that of the CT morphological feature model(P<0.05),and that there was no significant difference between the radiomics and comprehensive models(P>0.05).Subgroup analysis showed that the model performed consistently in the age and the sex subgroups(P>0.05).Conclusion The radiomics model was the most effective in predicting hypertension.When a patient is diagnosed with non-nodular adrenal hyperplasia by CT,the increase in adrenal morphology and texture heterogeneity may indicate an increase in the risk of hypertension.
范晨;陈维鹏;张晓玲;朱依谆
澳门科技大学中医药学院天然药物发现和产业化实验室,澳门 999078||珠海市人民医院医学影像中心,广东 珠海 519000珠海市人民医院医学影像中心,广东 珠海 519000珠海市人民医院医学影像中心,广东 珠海 519000澳门科技大学中医药学院天然药物发现和产业化实验室,澳门 999078
医药卫生
肾上腺非结节样增生高血压影像组学计算机断层扫描
non-nodular adrenal hyperplasiahypertensionradiomicscomputed tomography
《中国医科大学学报》 2026 (3)
227-233,7
澳门科学技术发展基金(FDCT0001/2024/AKP)
评论