基于BI-RADS 分类多参数乳腺MRI 对肿块性病变的诊断效能研究

Diagnostic value of multi-parameter BI-RADS classification MRI in breast lessions

摘要:

目的 评价乳腺MRI 乳腺影像报告和数据系统（BI-RADS）分类各观察指标对肿块性乳腺病变的 阳性预测价值（PPV）及诊断效能；探讨乳腺肿块性病变的动态增强磁共振（DCE-MRI）征象、功能磁共振 成像（DWI-ADC）值联合应用对该类良、恶性病变诊断的价值。方法 回顾性分析194 例乳腺肿块性病变， 计算BI-RADS 分类各指标的PPV，并分析MRI 常规检查（形态、边界、边缘及内部强化方式）、动态增强 扫描[ 时间- 信号曲线（TIC）] 和DWI 结果：以病理结果为“金标准”，分别得出PPV 和Kappa 值。结果 194 例肿块性病变中，BI-RADS 3 ～ 5 类PPV 分别为0.114、0.286、0.904 ；依据MRI BI-RADS 词汇形态学描述，形 态不规则PPV 为0.780，边缘不规则/ 毛刺PPV 为0.849，边缘强化PPV 为0.889 ；TIC 平台型和流出型PPV 分 别为0.613 和0.833 ；ADC 值≤ 1.18×10-3 mm2/s PPV 高达0.901。形状、病灶强化方式与病理结果吻合度较弱 kappa 值分别为0.410 和0.386，边界与病理结果吻合度检验水平一般kappa 值0.570。TIC 曲线及ADC 值与病理 结果吻合度检验水平较高kappa 值分别为0.735 和0.823。Logistic 回归分析结果表明ADC 值、边缘及TIC 类型 对乳腺良、恶性病变的诊断能力较准确（O^R = 01.041，3.156 和2.217），ADC 值≤ 1.18×10-3 mm2/s、边缘毛刺征、 TIC Ⅱ型和Ⅲ型为MRI 诊断恶性乳腺肿块性病变的危险因素（O^R = 01.041，3.156 和2.217）。结论 运用BIRADS 词汇中的标准术语对乳腺MRI 中肿块性病变特点进行严谨评估，对预测肿块性病变的恶性可能性有很 大作用。边缘、TIC 曲线及ADC 值在乳腺良、恶性病变诊断中具有良好的诊断效能。

Abstract:

Objective To evaluate the diagnostic value of breast MRI BI - RADS classification in breast mass. Method Totally 194 cases with breast mass lesions was involved in this study. Positive predictive value (PPV) of the BI-RADS classification index, MRI examination (edge shape, boundary, and the internal enhancement), dynamic enhanced scan (TIC curve) and DWI was performed and recorded. Statistical analysis was conducted for identification of predictive significance. Result PPV of breast mass as BI-RADS 3, 4, and 5 were 0.114, 0.286, and 0.904, respectively. PPV of irregular shape, irregular margin, marginal enhancement was 0.780, 0.849, and 0.889, respectively. The PPV of Platform type and washout type were 0.613 and 0.833 respectively. PPV of ADC values ≤ 1.18×10-3 mm2/s was as high as 0.901. Shape, Margin, andfeatures of enhancement did not match well with pathological results. TIC curve and ADC value matched satisfactorily with pathological results. Logistic regression analysis results showed that ADC values, irregular margin and TIC curve type were independent risk factors for diagnosis of malignant breast lesions, with O^R value as 1.041, 3.156, and 2.217, respectively. Conclusion BI-RADS may be able to predict malignancy of breast lesions. Margin, TIC curve and ADC values are independent risk factors for malignant breast lesions.