MRI与X线联合评估乳腺非肿块样病变恶性风险的Logistic回归模型建立及应用评价
Establishment and application evaluation of Logistic regression model for diagnosing malignant risk of breast non-mass-kike lesions based on MRI and mammography
姚远 1张海金 1张文婷 1刘辉 1卞巍1
作者信息
- 1. 嘉兴市妇幼保健院放射科,浙江 嘉兴 314000
- 折叠
摘要
目的:基于MRI与X线特征建立乳腺非肿块样病变(Non-mass-like lesions,NML)恶性风险的预测模型,期望提高诊断准确率,为临床决策提供依据.方法:对我院 2021 年 7 月—2023 年 4 月符合纳入标准的 150 例患者的资料进行回顾性分析,使用Logistic回归构建NML恶性风险预测模型及列线图,采用校准曲线评估模型准确度,用受试者工作特征(ROC)曲线评估模型的诊断效能.结果:多因素分析显示簇状环形强化、时间-信号强度曲线(TIC)类型、ADC值、线样及段样分布钙化等影像特征是预测病变恶性风险的因素.基于MRI特征的模型的ROC曲线下面积为 0.941,灵敏度为 88.7%,特异度为 86.6%.基于MRI联合X线特征的模型的ROC曲线下面积为 0.951,灵敏度为 91.5%,特异度为 91.4%,校准曲线预测准确度较好.结论:基于MRI联合X线特征建立的乳腺NML恶性风险Logistic回归模型诊断效能较高,具有一定的应用潜力.
Abstract
Objective:To develop,verify and test a predictive model for the malignant risk of breast non-mass-like le-sions(NML)based on magnetic resonance imaging(MRI)and mammography features,in order to improve the accuracy of malig-nant diagnosis and provide evidence for clinical decision-making.Methods:This was a retrospective study,which included the radiologic data of 150 cases that met the inclusion criteria from July 2021 to April 2023.Logistic regression was con-ducted to obtain the malignant risk prediction model of breast NML,as well as the nomogram.The accuracy of this model was evaluated by the calibration curve.The diagnostic performance was evaluated by the receiver operating characteristic(ROC)curve.Results:There were multiple factors related to the malignant risk of NML,including clustered ring enhancement,type of time-intensity curve(TIC),apparent diffusion coefficient(ADC)value,and linear or segmental calcification distribution.The model based on MRI features had an area under the ROC curve(AUC)of 0.941,a sensitivity of 88.7%,and a specificity of 86.6%.The model based on the combination of MRI and mammography features had an AUC of 0.951,a sensitivity of 91.5%,and a specificity of 91.4%.The calibration curve indicated good predictive accuracy.Conclusion:The model for the diagnosis of malignant risk of breast NML based on the combination of MRI and mammography features demonstrated high di-agnostic performance and had potential applications.
关键词
乳腺肿瘤/磁共振成像/放射摄影术Key words
Breast Neoplasms/Magnetic Resonance Imaging/Radiography引用本文复制引用
基金项目
嘉兴市科技计划项目(2021AD30028)
出版年
2024
国家科技期刊平台
NSTL
万方数据