基于县域医院场景对比自动乳腺超声（ABUS）远程阅片模式和手持超声（HHUS）乳腺癌的诊断效能，评估ABUS远程阅片模式在乳腺癌筛查中的临床性能。

回顾性纳入2021年4~10月于广东省龙川县妇幼保健医院进行乳腺检查的女性患者454例，所有患者均完成HHUS和ABUS检查，并分别给出美国放射学会乳腺成像报告和数据系统（BI-RADS）分类，HHUS的扫查与阅片由4名5年以上乳腺超声工作经验的超声医师完成，并记录消耗时间；ABUS图像采集由1名至少1年工作经验的技师完成全乳标准化超声图像采集，通过专用软件传输至远程阅片中心，由2名5年以上乳腺专科超声医师独立阅片，同时记录消耗时间。所有超声检查图像由阅片人员进行独立盲法判读。根据组织病理学及两年随访结果确定恶性、良性及正常病例，应用McNemar检验对比分析两种检查结果的敏感度、特异度、阳性预测值、阴性预测值，应用Delong检验比较受试者工作特征曲线（ROC）曲线下面积（AUC），应用配对样本t检验分析总耗时、超声医师耗时，使用χ²检验比较BI-RADS 1~2类、BI-RADS 3类和BI-RADS 4~5类占比。以P＜0.05为差异有统计学意义。

研究共纳入女性454例，年龄范围20~67岁，平均年龄（41.49±7.08）岁，其中恶性病例9例，良性或正常病例445例；ABUS检出恶性病例9例，HHUS检出恶性病例6例，两种检查方法在特异度、阳性预测值和AUC组间差异有统计学意义（96.9%、90.8%，P＜0.01；39.1%、12.8%，P＜0.01；0.99、0.85，P=0.028）。ABUS和HHUS两种检查方法的平均总耗时为（836±78）s、（311±77）s；超声医师平均耗时为（236±31）s、（311±77）s，组间差异有统计学意义（P均＜0.05）。BI-RADS1~2类在ABUS和HHUS两种检查方式中的占比分别为68.7%（312/454）、54.2%（246/454）；BI-RADS3类占比分别为26.2%（119/454）、35.5%（161/454）；BI-RADS4~5类占比分别为5.1%（23/454）、10.4%（47/454）；组间差异均有统计学意义（P＜0.05）。

在县域医院场景下，ABUS远程阅片模式诊断乳腺癌较HHUS具有更高的诊断效能，且ABUS检查超声医师诊断耗时短于HHUS检查。

To compare the diagnostic efficacy of the remote-reading mode of Automated Breast Ultrasound (ABUS) and Handheld Ultrasound (HHUS) for breast cancer in county-wide hospitals, and to evaluate the clinical performance of the ABUS remote-reading mode in breast cancer screening.

A total of 454 women who underwent breast examinations at the Maternal and Child Health Hospital of Longchuan County, Guangdong Province, from April to October 2021 were retrospectively included. All of them completed both HHUS and ABUS examinations and were assigned BI-RADS classifications accordingly. The scanning and image-reading of HHUS were performed by four ultrasound physicians with over 5 years of experience in breast ultrasound, and the time spent was recorded. ABUS image acquisition was conducted by a technician with at least one year of work experience. The whole-breast images were transmitted to a reading center via dedicated software. Two ultrasound physicians with more than 5 years of experience in breast-specific ultrasound independently interpreted the images and recorded the time spent. All ultrasound images were indepentty reviewed by the readers in a blinded manner. Malignant cases were confirmed by histopathology, while benign and normal cases were confirmed by pathology or two-year follow-up results. The McNemar's test was used to compare sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV); the DeLong's test was used for area under the curve (AUC) comparison; the paired-samples t-test was applied to compare total time and ultrasound physician time; and the chi-square test was used to compare the proportional distribution of BI-RADS 1~2, 3, and 4~5 categories.

A total of 454 women were included in our study, with a mean age of (41.49±7.08) years (range: 20~67 years). There were 9 malignant cases and 445 benign and normal cases. Nine malignant cases were detected by ABUS and six by HHUS. Statistically significant differences were found between the two examinations in specificity, positive predictive value, and the AUC (96.9% and 90.8%, P＜0.01; 39.1% and 12.8%, P＜0.01; 0.99 and 0.85, P=0.028, respectively). The average total time spent was (836±78) seconds for ABUS and (311±77) seconds for HHUS; the average ultrasound physician time spent was (236± 31) seconds for ABUS and (311±77) seconds for HHUS (all P＜0.05). For ABUS and HHUS, respectively, the proportions of BI-RADS 1~2 lesions were 68.7% (312/454) and 54.2% (246/454), BI-RADS 3 were 26.2% (119/454) and 35.5% (161/454), and BI-RADS 4~5 were 5.1% (23/454) and 10.4% (47/454). All intergroup differences were statistically significant (P＜0.05).

In county-wide hospital settings, the remote-reading mode of ABUS demonstrates higher diagnostic performance for breast cancer compared with HHUS, and the diagnostic time consumed by ultrasound physicians for ABUS is shorter than that for HHUS.