构建多模态磁共振评分模型鉴别肿块型乳腺病变良恶性  

A Multimodal Magnetic Resonance Imaging Scoring Model for Differentiation Between Benign and Malignant Breast Masses

作　　者：张刚[1] 刘永天[1] 周彦汝 沈智威 ZHANG Gang;LIU Yong-tian;ZHOU Yan-ru;SHEN Zhi-wei(Department of MRI,The First Affiliated Hospital of Henan University of CM,Zhengzhou 450000,Henan Province,China;Philips Health Technology,Beijing 100000,China)

机构地区：[1]河南中医药大学第一附属医院磁共振室,河南郑州450000 [2]飞利浦健康科技,北京100000

出　　处：《中国CT和MRI杂志》2023年第9期97-100,共4页Chinese Journal of CT and MRI

基　　金：2021年河南省卫生健康委国家中医临床研究基地科研专项(2021JDZY095)。

摘　　要：目的建立基于多模态MRI图像特征的简易评分模型,为肿块型乳腺病变良恶性鉴别诊断提供简便有效的工具。方法回顾性分析经病理检查证实的肿块型乳腺病变的MRI资料,其中40例来自良性肿块病变,51例来自恶性肿块病变。对比良恶性病变的影像学特征,找出具有显著性差异的指标,使用受试者工作特征(ROC)曲线,评价其诊断效能,最后采用多因素Logistic回归分析的方法,建立多模态MRI简易评分模型鉴别乳腺肿块良恶性。结果经单因素方差分析,筛选出具有组间统计学差异的指标(良恶性组肿块形状、边缘、最大径、早期强化比值、达峰时间、内部强化类型、TIC曲线、ADC值)。经ROC曲线分析,筛选出AUC值大于0.8的指标(边缘,ADC值和TIC曲线),应用多因素Logistic回归分析,构建了乳腺MRI良恶性肿块鉴别诊断回归模型。该模型具有很高的乳腺肿块良恶性鉴别诊断效能,其敏感性和特异性分别达到96%和96%。结论基于乳腺肿块ADC值、边缘、TIC曲线的的多模态MRI评分模型,可便捷有效地鉴别良恶性乳腺肿块,具有较大推广意义。Objective To establish a simple scoring model based on multimodal MRI image features to provide an effective tool for the differentiation between benign and malignant breast masses.Methods The MRI data of breast mass lesions were retrospectively analyzed,40 cases from benign mass lesions and 51 cases from malignant mass lesions confirmed by pathological examination.Imaging specialties were compared,and the diagnostic efficacy was evaluated with subject operating characteristic(ROC)curves.A multivariate logistic regression analysis was used to establish an MRI-based differential diagnosis model for breast masses.Results Using one-way ANOVA,the indicators with statistical differences between groups(mass shape,margin,maximum diameter,internal strengthening characteristics,early strengthening ratio,time to peak,TIC curve,and ADC value)were screened out.With ROC curve analysis,the indicators(margins,ADC values,and TIC curves)with AUC values greater than 0.8 were screened,and a regression model was established for the differentiation between benign and malignant breast masses by applying multi-factor logistic regression analysis.The model has high efficacy in the differential diagnosis of benign and malignant breast masses,with an AUC of 0.98 and sensitivities and specificities of 96%and 96%,respectively.Conclusion The scoring model based on ADC values,margins,and TIC curves of breast MRI masses can conveniently and effectively identify benign and malignant breast masses and is easy to achieve clinical promotion.

关 键 词：乳腺 良恶性 磁共振成像 简易评分模型 

分 类 号：R445.2[医药卫生—影像医学与核医学]