Prospective implementation of AI-assisted screen reading to improve early detection of breast cancer

• Published in: Nature medicine Vol. 29; no. 12; pp. 3044 - 3049
• Main Authors: Ng, Annie Y., Oberije, Cary J. G., Ambrózay, Éva, Szabó, Endre, Serfőző, Orsolya, Karpati, Edit, Fox, Georgia, Glocker, Ben, Morris, Elizabeth A., Forrai, Gábor, Kecskemethy, Peter D.
• Format: Journal Article
• Language: English
• Published: New York Nature Publishing Group US 01-12-2023; Nature Publishing Group
• Subjects: 692/699/67/1347; 692/700/139; 692/700/1421/1770; 692/700/478/2772; Arbitration; Artificial Intelligence; Biomedical and Life Sciences; Biomedicine; Breast cancer; Breast Neoplasms - diagnosis; Cancer Research; Cancer screening; Early Detection of Cancer - methods; Female; Humans; Infectious Diseases; Mammography; Mammography - methods; Medical screening; Metabolic Diseases; Molecular Medicine; Neurosciences; Observer Variation; Prospective Studies; Reading; Retrospective Studies; Workflow
• ISSN: 1078-8956; 1546-170X; 1546-170X
• DOI: 10.1038/s41591-023-02625-9

Artificial intelligence (AI) has the potential to improve breast cancer screening; however, prospective evidence of the safe implementation of AI into real clinical practice is limited. A commercially available AI system was implemented as an additional reader to standard double reading to flag cases for further arbitration review among screened women. Performance was assessed prospectively in three phases: a single-center pilot rollout, a wider multicenter pilot rollout and a full live rollout. The results showed that, compared to double reading, implementing the AI-assisted additional-reader process could achieve 0.7–1.6 additional cancer detection per 1,000 cases, with 0.16–0.30% additional recalls, 0–0.23% unnecessary recalls and a 0.1–1.9% increase in positive predictive value (PPV) after 7–11% additional human reads of AI-flagged cases (equating to 4–6% additional overall reading workload). The majority of cancerous cases detected by the AI-assisted additional-reader process were invasive (83.3%) and small-sized (≤10 mm, 47.0%). This evaluation suggests that using AI as an additional reader can improve the early detection of breast cancer with relevant prognostic features, with minimal to no unnecessary recalls. Although the AI-assisted additional-reader workflow requires additional reads, the higher PPV suggests that it can increase screening effectiveness.
In a phased prospective rollout, the implementation of AI as an additional reader for mammography screening improved the real-world early detection of breast cancer compared to standard double reading involving two independent radiologists.