摘要
目的 本研究旨在探讨乳腺MRI对BI-RADS 4A和4B分类可疑病变的降级价值.方法 接受乳腺MRI检查的可疑病变被分为4A和4B两类的患者被纳入这项回顾研究,两名专科医生对MR图像进行评估.在动态增强MRI(DCE-MRI)和弥散加权成像(DWI)的基础上,协商一致地给出了MRI BI-RADS分类.以病理结果和影像随访至少12个月作为参考标准.计算敏感度、特异度、阳性预测值(PPV)、NPV及其95％可信区间(CI).结果 167名患者186个病变(US 4A类:145个,US4B类:41个)组成了研究队列,癌变率为34.9％(65/186).MRI显示所有恶性肿瘤均为真阳性,92.6％(112/121)的良性病变被正确诊断.MRI使PPV由34.9％(65/186)增加到87.8％(65/74),假阳性率降低92.6％(112/121).MRI的敏感性、特异性、PPV和NPV分别为 100％(95％CI:94.5％～100％)、92.6％(95％CI:86.3％～96.5％)、87.8％(95％CI:78.2％～94.3％)和 100％(95％CI:96.8％～100％).2.2％(4/186)的可疑病灶在MRI上被额外发现,其中75％(3/4)为恶性病变.结论 MRI可以降低BI-RADS 4A和4B类可疑病变的级别,避免不必要的良性活检,而不会漏诊恶性病变.MRI上发现的其他可疑病变需要进一步检查.
Abstract
Objective This study was to investigate the value of breast MRI in the degradation of suspicious lesions classified as BI-RADS 4A and 4B.Methods Patients who underwent breast MRI with suspicious lesions classified as 4A and 4B were included in this review study,and MR Images were evaluated by two specialists.On the basis of dynamic enhanced MRI(DCE-MRI)and diffusion-weighted imaging(DWI),the classification of MRI BI-RADS is given by consensus.Pathological findings and imaging follow-up for at least 12 months were used as reference standards.Sensitivity,specificity,positive predictive value(PPV),NPV and 95％confidence interval(CI)were calculated.Results 167 patients with 186 lesions(US 4A:145,US 4B:41)comprised the study cohort,with a cancer rate of 34.9％(65/186).MRI showed true positives for all malignancies,and 92.6％(112/121)of benign lesions were correctly diagnosed.MRI increased PPV from 34.9％(65/186)to 87.8％(65/74),and reduced the false positive rate by 92.6％(112/121).MRI of the sensitivity,specificity,PPV and NPV of 100％(95％CI:94.5％～100％),92.6％(95％CI:86.3％～96.5％),87.8％(95％CI:78.2％～94.3％)and 100％(95％CI:96.8％～100％).2.2％(4/186)of the suspected lesions were detected on MRI,of which 75％(3/4)were malignant lesions.Conclusion MRI can reduce the grade of suspected lesions of BI-RADS 4A and 4B,avoid unnecessary benign biopsy,and will not miss malignant lesions.Other suspicious lesions found on the MRI require further examination.
维普
万方数据