In a Swedish trial (MASAI) reported in The Lancet, Gommers et al found that artificial intelligence (AI)-supported mammography was noninferior to standard double reading without AI in identifying interval breast cancers in women undergoing breast cancer screening.
Study Details
In the study, 105,915 eligible women were randomly assigned between April 2021 and December 2022 to receive AI-supported mammography screening (n = 53,043) or standard double reading without AI (n = 52,872). In Sweden, eligible women are invited to screen every 1.5 to 2 years, or annually for those with higher risk. AI was used to triage examinations to single or double reading by radiologists and for detection support. Interval cancers were defined as those diagnosed between two screening rounds or within 2 years after the last scheduled screening that were not detected at screening. The primary outcome measure was interval cancer rate, with a noninferiority margin of 20%.
Key Findings
Interval cancer rates per 1,000 participants were 1.55 (95% confidence interval [CI] = 1.23–1.92) in the AI group vs 1.76 (95% CI = 1.42–2.15) in the control group, yielding a noninferior proportion ratio of 0.88 (95% CI = 0.65–1.18, P = .41). The absolute decrease in interval cancer rate in the AI group was 0.21 (95% CI = –0.28 to 0.70) per 1,000 participants. The AI group had lower numbers of participants with interval cancers that were invasive (75 vs 89), T2+ (38 vs 48), or nonluminal A (43 vs 59).
Sensitivity was 80.5% (95% CI = 76.4%–84.2%) in the AI group vs 73.8% (95% CI = 68.9%–78.3%) in the control group (P = .031). Specificity was 98.5% (95% CI = 98.4%–98.6%) in both groups (P = .88).
The investigators concluded: “AI-supported mammography screening showed consistently favourable outcomes compared with standard double reading, with a non-inferior interval cancer rate, fewer interval cancers with unfavourable characteristics, higher sensitivity, and the same specificity, while also reducing screen reading workload. These findings imply that AI-supported mammography screening can efficiently improve screening performance compared with standard double reading and may be considered for implementation in clinical practice.”
Kristina Lång, PhD, of Division of Diagnostic Radiology, Department of Translational Medicine, Lund University, Malmö, Sweden, is the corresponding author for The Lancet article.
DISCLOSURE: The study was funded by the Swedish Cancer Society and others. For full disclosures of the study authors, visit thelancet.com.
