自动乳腺全容积成像预测模型与增强MRI对乳腺癌诊断的价值研究

Comparative study of automated breast volume scan predictive model and dynamic contrast-enhanced MRI in diagnosis of breast lesions

目的分析自动乳腺全容积成像(ABVS)的图像特点,构建乳腺癌ABVS预测模型,与动态增强磁共振(DCE-MRI)对比该模型的诊断价值。方法选取乳腺病变患者212例,共215个病灶。随机分为两组:其中130例133个病灶经ABVS检查为ABVS组;另82例82个病灶经ABVS和DCE-MRI检查为ABVS+DCE-MRI组。对ABVS组图像特征进行Logistic回归分析,建立预测模型;评价对乳腺病灶的诊断效能。结果筛选出与乳腺癌相关性较高的图像特征为汇聚征(OR=12.899,P<0.05),成角(OR=7.095,P<0.05),微钙化(OR=5.980,P<0.05),建立的风险预测模型Y=-0.732+2.557X 2+1.788X 1+1.959X 4,该模型预测乳腺恶性结节的曲线下面积为0.849。乳腺ABVS预测模型与DCE-MRI诊断乳腺病灶的一致性高(Kappa值=0.891)。结论乳腺ABVS预测模型具有较高的诊断效能,与DCE-MRI诊断一致性好,可作为乳腺疾病诊断的一种新的有效的方法。

Objective To find enhanced features of diagnostic value of the automated breast volume scanner(ABVS),to build breast(ABVS)predictive models,and to compare the diagnostic value of this model with dynamic contrast enhanced magnetic resonance(DCE-MRI).Methods 212 breast patients with 215 lesions were selected in the study.In ABVS group,130 cases with 133 lesions were examined by ABVS,so as to find risk factors of ABVS pattern in breast malignant lesions by Logistic regression analysis,to build breast ABVS predictive models,and to draw ROC Curve.In ABVS+DCE-MRI group,82 cases with 82 lesions were examined by ABVS and DCE-MRI.We compared the diagnostic efficacy of ABVS prediction model and DCE-MRI in breast lesions with final pathology results as the gold standard.Results Three independent variables,namely,convergence sign(OR=12.899,P<0.05),angulation(OR=7.095,P<0.05)and microcalcification(OR=5.980,P<0.05)were selected in the final step of the logistic regression analysis in ABVS.The logistic regression equation was as follows:Y=-0.732+2.557X 2+1.788X 1+1.959X 4,breast ABVS predictive models area under ROC curve of benign and malignant breast lesions to be 0.849.Breast ABVS prediction model was highly consistent with DCE-MRI in the diagnosis of breast lesions(Kappa=0.891).Conclusion Breast ABVS prediction model has high diagnostic efficiency and good consistency with DCE-MRI.It can be used as a new and effective method for the diagnosis of breast diseases.