From the Department of Radiology and Nuclear Medicine, Radboud University Medical Center, PO Box 9101, 6500 HB Nijmegen, Geert Grooteplein 10, 6525 GA, Post 766, Nijmegen, the Netherlands (A.R.R., I.S., R.M.M.); Department of Radiology & Imaging Sciences, Emory University, Atlanta, Ga (E.K.); ScreenPoint Medical BV, Nijmegen, the Netherlands (J.J.M.); Lynn Women's Health & Wellness Institute, Boca Raton Regional Hospital, Boca Raton, Fla (K.S.); Referenzzentrum Mammographie Munich, Brustdiagnostik München and FFB, Munich, Germany (S.H.H.); and Dutch Expert Centre for Screening, Nijmegen, the Netherlands (I.S.).
Radiology. 2019 Feb;290(2):305-314. doi: 10.1148/radiol.2018181371. Epub 2018 Nov 20.
Purpose To compare breast cancer detection performance of radiologists reading mammographic examinations unaided versus supported by an artificial intelligence (AI) system. Materials and Methods An enriched retrospective, fully crossed, multireader, multicase, HIPAA-compliant study was performed. Screening digital mammographic examinations from 240 women (median age, 62 years; range, 39-89 years) performed between 2013 and 2017 were included. The 240 examinations (100 showing cancers, 40 leading to false-positive recalls, 100 normal) were interpreted by 14 Mammography Quality Standards Act-qualified radiologists, once with and once without AI support. The readers provided a Breast Imaging Reporting and Data System score and probability of malignancy. AI support provided radiologists with interactive decision support (clicking on a breast region yields a local cancer likelihood score), traditional lesion markers for computer-detected abnormalities, and an examination-based cancer likelihood score. The area under the receiver operating characteristic curve (AUC), specificity and sensitivity, and reading time were compared between conditions by using mixed-models analysis dof variance and generalized linear models for multiple repeated measurements. Results On average, the AUC was higher with AI support than with unaided reading (0.89 vs 0.87, respectively; P = .002). Sensitivity increased with AI support (86% [86 of 100] vs 83% [83 of 100]; P = .046), whereas specificity trended toward improvement (79% [111 of 140]) vs 77% [108 of 140]; P = .06). Reading time per case was similar (unaided, 146 seconds; supported by AI, 149 seconds; P = .15). The AUC with the AI system alone was similar to the average AUC of the radiologists (0.89 vs 0.87). Conclusion Radiologists improved their cancer detection at mammography when using an artificial intelligence system for support, without requiring additional reading time. Published under a CC BY 4.0 license. See also the editorial by Bahl in this issue.
目的 比较放射科医师在未使用和使用人工智能（AI）系统支持的情况下对乳腺 X 线摄影检查进行乳腺癌检测的性能。
材料与方法 进行了一项丰富的回顾性、完全交叉、多读者、多病例、符合 HIPAA 标准的研究。纳入了 2013 年至 2017 年期间进行的 240 名女性（中位年龄，62 岁；范围，39-89 岁）的筛查性数字乳腺 X 线摄影检查。这 240 次检查（100 次显示癌症，40 次导致假阳性召回，100 次正常）由 14 名符合 Mammography Quality Standards Act 标准的放射科医师进行解读，一次是在有 AI 支持的情况下，一次是在没有 AI 支持的情况下。读者提供了乳腺成像报告和数据系统评分以及恶性肿瘤的可能性。AI 支持为放射科医师提供了交互式决策支持（点击乳房区域可获得局部癌症可能性评分）、用于计算机检测异常的传统病变标志物以及基于检查的癌症可能性评分。使用混合模型方差分析和广义线性模型对多次重复测量进行分析，比较了在有和没有 AI 支持的情况下的曲线下面积（AUC）、特异性和敏感性以及阅读时间。
结果 平均而言，使用 AI 支持时的 AUC 高于未使用 AI 支持时（分别为 0.89 和 0.87；P =.002）。使用 AI 支持时，敏感性提高（86% [86/100] 比 83% [83/100]；P =.046），而特异性有改善趋势（79% [111/140] 比 77% [108/140]；P =.06）。每个病例的阅读时间相似（未使用 AI 支持时为 146 秒，使用 AI 支持时为 149 秒；P =.15）。仅使用 AI 系统的 AUC 与放射科医师的平均 AUC 相似（0.89 比 0.87）。
结论 放射科医师在使用 AI 系统进行支持时提高了乳腺 X 线摄影检查的癌症检出率，而无需额外的阅读时间。
在 CC BY 4.0 许可下发布。请参阅本期 Bahl 编辑的社论。
