使用乳腺影像报告和数据系统诊断乳腺疾病的体会

The application experiences of breast imaging reporting and data system in mammographic diagnoses of breast lesion with symptoms

目的 探讨乳腺影像报告和数据系统 (BI RADS)在有症状乳腺病变X线诊断中的使用价值。方法 高、低年资两组医生回顾分析 30 7例乳腺病变的X线片并作BI RADS分级 ,用操作者特征曲线 (ROC)、Kappa分析两组医生判读情况 ,并与病理对照分析。 结果 30 7例中乳腺癌15 9例 ,良性病变 14 8例。高、低年资两组医生对乳腺病变的BI RADS判读分级ROC曲线下面积(Az)分别为Az1=0 94 6 ,Az2 =0 84 5 ,U检验差异无显著性意义 (U =0 5 2 6 ,P >0 0 5 ) ,总符合率有中至高度一致性 ,K =0 4 35 ,其中Ⅰ级 (0 5 11)、Ⅴ级 (0 6 79)为中至高度符合 ,Ⅱ级 (0 373)、Ⅲ级(0 179)、Ⅳ级 (0 2 76 )符合较差。高年资组与低年资组各分级中乳腺癌的阳性率分别为Ⅰ级(12 5 %、2 2 5 % )、Ⅱ级 (3 7%、12 5 % )、Ⅲ级 (2 0 83%、19 6 1% )、Ⅳ级 (6 8 6 3%、6 0 34% )和Ⅴ级(10 0 %、89 91% )。两组医生判读均错误 2 3例 (7 4 9% )。结论 BI RADS在有症状乳腺病变的X线诊断中有价值。合理应用 0级、结合临床检查能降低漏诊率。对有症状的乳腺病变 ,Ⅱ、Ⅲ级诊断域值可提高 ,Ⅳ级域值可降低。BI RADS有其局限性 。

Objective To study the use of breast imaging reporting and data system (BI-RADS) in final assessment among observers and to evaluate its value in breast lesion with symptom. Methods Four doctors divide into two groups (senior and junior groups) assessed and classified mammograms in (307 patients) according to BI-RADS (category 0-category Ⅴ), and further ameliorate the use in symptomatic patients in comparison with the pathologic findings. The method was tested using ROC and Kappa. Results(The lesions) included 159 cancers and 148 benign lesions. The areas under the ROC curves in senior and junior doctor groups were Az_1=0.946 and Az_2=0.845 respectively. There was no statistical differences in BI-RADS category between the two groups (U=0.526, P>0.05). A Kappa value was 0.435 for overall final assessment. There were moderate agreement in category Ⅰ and Ⅴ (K=0.511 and 0.671), and slight agreement in category Ⅱ (0.373), Ⅲ (0.179), and Ⅳ (0.276). The positive value of different categories assessed by two groups were as follows: category Ⅰ (12.5% vs 22.5%), Ⅱ(3.7% vs 12.5%), Ⅲ(20.83% vs 19.61%), Ⅳ(68.63% vs 60.34%), and Ⅴ(100% vs 89.91%). Twenty three lesions(7.49%)were misdiagnosed by both groups. Conclusion BI-RADS is useful in mammographic diagnoses of breast lesion with symptoms. The misdiagnoses can be reduced combined with the clinical examination and reasonable additional imaging evaluation. For symptomatic breast lesions, the assessing threshold of category Ⅱ or Ⅲ should be improved, and category Ⅳ should be reduced. But BI-RADS has its limitation. We propose that corresponded classification should be drawn according to every application’s condition.