乳腺影像报告与数据系统在鉴别诊断乳腺良恶性肿瘤中的应用价值

Value of Breast Imaging Reporting and Data System in the Differential Diagnosis of Benign and Malignant Breast Carcinomas

目的评价乳腺影像报告与数据系统(BI-RADS)在乳腺良恶性肿瘤鉴别诊断中的应用价值。资料与方法根据BI-RADS规范,分析202例患者共234个肿瘤的形态、方向、边缘、边界、内部回声、后方回声及钙化情况,并将病灶定义为3级(可能良性)、4级(可能恶性)和5级(高度可疑恶性)。计算每个病灶特征的阳性预测值与阴性预测值,分析影响诊断的因素。结果 49个(20.9%)为BI-RADS3级,47个(20.1%)为BI-RADS4级,138个(59.0%)为BI-RADS5级,BI-RADS3级对乳腺肿瘤的阴性预测值为98.0%(48/49),BI-RADS4级与5级对乳腺肿瘤的阳性预测值分别为34.0%(16/47)和92.8%(128/138)。BI-RADS对乳腺恶性肿瘤的诊断敏感性为99.3%,特异性为53.9%,阳性预测值为77.8%,阴性预测值为98.0%。对恶性肿瘤的诊断有较高阳性预测值的超声特征包括不完整的边缘(100.0%)、微钙化(100.0%)、不清晰的边界(97.3%)、后方回声衰减(97.0%);对良性肿瘤的诊断有较高阳性预测值的超声特征包括后方回声增强(100.0%)、形态规则(98.0%)、病灶方向平行于皮肤长轴(98.0%)、边缘完整(98.0%)、边界清晰(98.0%)与无钙化(98.0%)。结论 BI-RADS可帮助分析影响诊断的因素,有助于乳腺良恶性肿瘤的鉴别诊断。

Purpose To assess the applicability of the sonographic breast imaging reporting and data system （BI-RADS） classification for differentiating benign from malignant breast lesions. Materials and Methods A total of 202 patients with 234 breast lesions were included in this study. Each lesion was analyzed according to BI- RADS lexicon for ultrasonic features including shape, orientation, margin, boundary, internal echo pattern, posterior acoustic feature and calcification, and was categorized as category 3 （probably benign）, category 4 （probably malignant）, or category 5 （highly suggestive of malignancy）. Positive predictive value and negative predictive value for each sonographic descriptor were calculated. The factors affecting diagnosis were also analyzed. Results BI-RADS assessment categories were category 3 in 49 lesions （20.9%）, category 4 in 47 （20.1%）, and category 5 in 138 （59.0%）. The negative predictive value of category 3 for the diagnosis of breast lesions was 98.0%,5 （48/49）, and the positive predictive values of category 4 and 5 for the diagnosis of breast lesions were 34.0% （16/47） and 92.8% （128/138）, respectively. The sensitivity, specificity, positive predictive value and negative predictive value of BI-RADS category for the diagnosis of malignancy were 99.3%, 53.9%, 77.8%, and 98.0%, respectively. The features with highest positive predictive value for malignancy were noncircumscribed margin （100.0%）, microcalcification （100.0%）, indistinct boundary （97.3%）, and posterior acoustic shadow （97.0%）. BI-RADS descriptors showing high positive predictive value for benign lesions included posterior acoustic enhancement （100.0%）, regular shape （98.0%）, parallel orientation （98.0%）, circumscribed margin （98.0%）, abrupt interface （98.0%） and noncalcification （98.0%）. Conclusion The BI-RADS classification is helpful in evaluating diagnosis and differentiating benign from malignant breast mass.