From the Department of Radiology, Severance Hospital, Research Institute of Radiological Science, Yonsei University, College of Medicine, 50 Yonsei-ro, Seodaemun-gu, 03722 Seoul, Korea (J.H.Y.); Department of Oncology and Pathology, Karolinska Institute, Stockholm, Sweden (F.S.); Department of Radiology, Unit of Breast Imaging, Karolinska University Hospital, Stockholm, Sweden (F.S.); Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria (P.A.T.B.); Department of Radiology, University of Pennsylvania, Philadelphia, Pa (E.F.C.); Department of Radiology, University of Cambridge, Cambridge, UK (F.J.G.); Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Boston, Mass (C.D.L.); Department of Radiology, University of California Davis, Davis, Calif (E.A.M.); Department of Radiology, Breast Imaging Division, Johns Hopkins Medicine, Baltimore, Md (L.A.M.); Department of Radiology, University of Pittsburgh, UPMC Magee-Womens Hospital, Pittsburgh, Pa (R.M.N.); Department of Radiology, St James Hospital, Leeds, UK (N.S.); Department of Breast Examinations, Copenhagen University Hospital Herlev-Gentofte, Copenhagen, Denmark (I.V.); Department of Radiology, Laura and Isaac Perlmutter Cancer Center, Center for Biomedical Imaging, Center for Advanced Imaging Innovation and Research, New York University Grossman School of Medicine, New York, NY (L.M.); Department of Medical Imaging, Radboud University Medical Center, Nijmegen, the Netherlands (R.M.M.); and Department of Radiology, Netherlands Cancer Institute, Amsterdam, the Netherlands (R.M.M.).
Radiology. 2023 Jun;307(5):e222639. doi: 10.1148/radiol.222639. Epub 2023 May 23.
Background There is considerable interest in the potential use of artificial intelligence (AI) systems in mammographic screening. However, it is essential to critically evaluate the performance of AI before it can become a modality used for independent mammographic interpretation. Purpose To evaluate the reported standalone performances of AI for interpretation of digital mammography and digital breast tomosynthesis (DBT). Materials and Methods A systematic search was conducted in PubMed, Google Scholar, Embase (Ovid), and Web of Science databases for studies published from January 2017 to June 2022. Sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) values were reviewed. Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 and Comparative (QUADAS-2 and QUADAS-C, respectively). A random effects meta-analysis and meta-regression analysis were performed for overall studies and for different study types (reader studies vs historic cohort studies) and imaging techniques (digital mammography vs DBT). Results In total, 16 studies that include 1 108 328 examinations in 497 091 women were analyzed (six reader studies, seven historic cohort studies on digital mammography, and four studies on DBT). Pooled AUCs were significantly higher for standalone AI than radiologists in the six reader studies on digital mammography (0.87 vs 0.81, = .002), but not for historic cohort studies (0.89 vs 0.96, = .152). Four studies on DBT showed significantly higher AUCs in AI compared with radiologists (0.90 vs 0.79, < .001). Higher sensitivity and lower specificity were seen for standalone AI compared with radiologists. Conclusion Standalone AI for screening digital mammography performed as well as or better than radiologists. Compared with digital mammography, there is an insufficient number of studies to assess the performance of AI systems in the interpretation of DBT screening examinations. © RSNA, 2023 See also the editorial by Scaranelo in this issue.
背景 人工智能（AI）系统在乳腺筛查中的潜在应用引起了广泛关注。然而，在 AI 能够成为独立的乳腺影像学解释手段之前，对其性能进行严格评估是至关重要的。
目的 评估 AI 用于解读数字乳腺摄影和数字乳腺断层合成（DBT）的独立性能。
材料与方法 系统检索了 PubMed、Google Scholar、Embase（Ovid）和 Web of Science 数据库中 2017 年 1 月至 2022 年 6 月发表的研究。审查了敏感性、特异性和受试者工作特征曲线下的面积（AUC）值。使用诊断准确性研究的质量评估 2（QUADAS-2）和比较（QUADAS-C）分别评估研究质量。对总体研究以及不同研究类型（阅读器研究与历史队列研究）和成像技术（数字乳腺摄影与 DBT）进行了随机效应荟萃分析和荟萃回归分析。
结果 共分析了 16 项研究，这些研究包括 497091 名女性的 1108328 次检查（6 项阅读器研究、7 项数字乳腺摄影的历史队列研究和 4 项 DBT 研究）。在 6 项数字乳腺摄影的阅读器研究中，与放射科医生相比，独立 AI 的 AUC 明显更高（0.87 比 0.81， =.002），但在历史队列研究中并非如此（0.89 比 0.96， =.152）。4 项 DBT 研究显示，与放射科医生相比，AI 的 AUC 明显更高（0.90 比 0.79， <.001）。与放射科医生相比，独立 AI 的敏感性更高，特异性更低。
结论 用于筛查数字乳腺摄影的独立 AI 的性能与放射科医生相当或更好。与数字乳腺摄影相比，用于评估 AI 系统在 DBT 筛查检查中的性能的研究数量不足。
© 2023 RSNA，见本期 Scaranelo 社论。
