对比分析数字化乳腺断面合成技术与全视野数字化乳腺X线摄影对不同乳腺类型肿块的诊断

Comparative Analysis of Digital Breast Tomosynthesis and Full Field Digital Mammography in Diagnosing Different Types of Breast Masses

目的对比分析数字化乳腺断面合成技术(DBT)与全视野数字化乳腺X线摄影(FFDM)对不同乳腺类型肿块的诊断及边缘特征。资料与方法回顾性分析2015年6_12月在上海市徐汇区大华医院同时行乳腺DBT与FFDM检查的189例患者,其中发现并证实肿块145例,比较2种检查方法下乳腺肿块的检出率,对比分析肿块晕征、遮蔽情况、微分叶、毛刺状情况。以病理结果及24个月随访结果为依据,结合FFDM和DBT 2种方法下的美国放射协会乳腺影像报告和数据系统(BI-RADS)分类,采用受试者工作特性(ROC)曲线分析2种检查方法的诊断效能。结果 189例患者中,发现非致密型乳腺肿块33例,致密型乳腺肿块112例;良性122例,恶性23例。非致密型乳腺肿块中,DBT与FFDM对肿块检出率分别为90.9%(30/33)、84.8%(28/33),差异有统计学意义(x^2=1.705,P=0.192),肿块晕征、遮蔽情况、微分叶、毛刺状数目及诊断效能比较,差异均无统计学意义(P均>0.05)。致密型乳腺肿块中,DBT与FFDM对肿块检出率分别为86.6%(97/112)、65.1%(73/112),差异有统计学意义(x^2=13.268,P<0.01);肿块的晕征[41.1%(30/73)比16.4%(12/73),x^2=10.830,P=0.001]、边缘遮蔽情况[38.4%(28/73)比60.3%(44/73),x^2=7.015,P=0.008]比较,差异均有统计学意义。DBT诊断非致密型乳腺肿块的ROC曲线下面积为0.962,FFDM为0.943(Z=0.028,P=0.978);DBT诊断致密型乳腺肿块的ROC曲线下面积为0.967,FFDM为0.831,差异有统计学意义(Z=2.267,P=0.024)。结论在致密型乳腺中,乳腺DBT较FFDM对肿块检出率、晕征、减少肿块遮蔽及诊断效能上均具有优势,可提高对乳腺肿块BI-RADS分类的准确率。

Purpose To compare and analyze the diagnosis and marginal features of different type breast masses by digital breast tomosynthesis （DBT） and full field digital mammography （FFDM）. Materials and Methods One hundred and eighty-nine patients who underwent DBT and FFDM examinations at Dahua Hospital in Xulmi District, Shanghai from June to December 2015 were retrospectively analyzed, among which 145 patients were found and confirmed to have masses. Detection rate of breast masses under two examination methods was counted and also tumor halo sign, obscuration, differential leaf and burr were compared and explored. The receiver operating characteristic （ROC） curve was used to analyze diagnostic efficacy of two methods based on pathological and 24-month follow-up results and combined with American Radiological Association Breast Imaging Reporting and Data System （BI-RADS） classification under FFDM and DBT methods. Results Of the 189 patients, 33 were with non-dense breast masses and 112 were with dense breast masses. 122 were benign, and 23 were malignant. In the non-dense breast masses, detection rates of masses with DBT and FFDM were 90.9% （30/33） and 84.8% （28/33）, respectively （χ^2=1.705, P=0.192）. There was no statistical significance in the comparison of tumor halo sign, obscuration, differential leaf, number of burrs and diagnostic efficacy （P〉0.05）. In the dense breast mass, detection rates of masses with DBT and FFDM were 86.6% （97/112） and 65.1% （73/112）, respectively （χ^2=13.268, P〈0.01）. Halo sign of the masses [41.1% （30/73）： 16.4% （12/73）, Z2=10.830, P=-0.001], edge obscuration [38.4% （28/73） ： 60.3% （44/73）, χ^2=7.015, P=-0.008] were compared and the differences were of statistical significance. The area under ROC curve for non-dense breast masses diagnosed with DBT and FFDM was 0.962 and 0.943, retrospectively （Z=0.028, P=0.978）; the area under the ROC curve for DBT diagnosis of dense breast mass was 0.967 and FFDM was 0.831 （Z=2.267, P=0.024）. Conclusion Breast DBT technology was advantageous in masses detection rate, halo signs, reducing masses obscuration and diagnosis efficacy compared with FFDM in dense breast and can improve accuracy of BI-RADS classification of breast masses.