乳腺肿块超声BI-RADS描述词的一致性及临床意义

Consistency of Describing Breast Mass by the Breast Imaging Reporting and Data System Ultrasound Lexicon and Its Clinical Significance

目的探讨乳腺肿块超声乳腺影像报告和数据系统(BI-RADS)描述词评估者间的一致性及其临床意义。资料与方法 154个乳腺肿块按照大小分为大肿块组(直径>1.0 cm,82个)和小肿块组(直径≤1.0 cm,72个),参照超声BI-RADS影像词典,由2名评估者采用双盲法对肿块形态、方位、边缘、边界、后方回声及微钙化进行观察、判读,与手术病理结果进行对照,比较各描述词评估者间的一致性及两组肿块超声BI-RADS的诊断效能。结果 2名评估者对大肿块组的形态、方位、边缘、边界和微钙化判断的一致性较好(Kappa=0.780、0.909、0.791、0.637、0.618,P<0.05);后方回声判断一致性中等(Kappa=0.535,P<0.05),对小肿块组的形态、方位、微钙化的判断一致性好(Kappa=0.704、0.621、0.800,P<0.05);边缘和后方回声的判断一致性中等(Kappa=0.421、0.460,P<0.05);边界的判断一致性差(Kappa=0.386,P<0.05)。与术后病理结果对照,大肿块组诊断准确度、敏感度、特异度分别为85.4%、96.4%、63.0%,小肿块组分别为68.1%、80.0%、56.7%,两组准确度和敏感度差异有统计学意义(P<0.05),特异度差异无统计学意义(P>0.05)。结论 BI-RADS描述病灶特征术语比较详细,病灶大小对判断结果有一定的影响。

Purpose To investigate the consistency of describing breast mass by the breast imaging reporting and data system （B1-RADS） ultrasound （US） lexicon and its clinical significance. Materials and Methods One hundred and filly-four breast tumors were divided into large tumor group （diameter〉1.0 cm, 82 masses） and small tumor group （diameter _〈 1.0 cm, 72 masses） according to the tumor size. Referring to the BI-RADS- US lexicon, shape, position, edge, border, posterior echo and microcalcification of the tumors were selected as observation indicators, and compared with pathology. The degree of intraobserver agreement was evaluated and BI-RADS-US diagnostic efficacy between two groups was compared. Results The greatest degree of reliability between the two evaluators was found for shape, orientation, margin, border and microcalcification in large tumor group （Kappa=0.780, 0.909, 0.791, 0.637, 0.618, P〈0.05）, but moderate agreement tbr posterior echo （Kappa=0.535, P〈0.05）. Two raters had good consistency for shape, orientation, and microcalcification in small tumor group （Kappa=0.704, 0.621, 0.800, P〈0.05）, moderate consistency for edge and posterior echo （Kappa-0.421, 0.460, P〈0.05）, but poor consistency for border （Kappa=0.386, P〈0.05）, Two evaluators reviewed all ultrasonogram again, and made coincidental BI-RADS-US classification for disagreement cases, which was compared with postoperative pathological findings. Diagnostic accuracy, sensitivity and specificity of large tumor group were 85.4% 96.4%, 63.0%0, and those of small tumor group were 68.1%, 80.0%, 56.7%, respectively. Tile accuracy and sensitivity between two groups had statistical significance （P〈0.05）, while specificity had no significant difference （P〉0.05）. Conclusion BI-RADS terminologies describe the lesion in detailed. Lesion size has a certain impact on final results.