AI Correctly Localized Almost One-Third of Interval Cancers on Screening Digital Breast Tomosynthesis

THURSDAY, July 31 Aug. 1, 2025 (HealthDay News) — Artificial intelligence (AI) correctly localized nearly one-third of interval cancers on retrospective evaluation of screening digital breast tomosynthesis (DBT) examinations, according to a study published online July 29 in Radiology.

Manisha Bahl, M.D., M.P.H., from Massachusetts General Hospital in Boston, and colleagues examined the performance of an AI algorithm in detecting and localizing interval cancers at screening DBT and validated the diagnostic threshold of the algorithm. Screening DBT examinations performed immediately before confirmed interval cancer diagnoses were retrospectively analyzed by a U.S. Food and Drug Administration-cleared AI algorithm. Lesions marked by AI on DBT sections were assigned a score from 0 to 100. Two breast imaging radiologists independently reviewed AI-positive examinations (score ≥10) to examine whether AI annotations corresponded with the site of subsequent cancer.

The researchers found that AI correctly localized 32.6 percent of 224 interval cancers in 224 women at retrospective evaluation of the screening DBT examinations. Larger size at surgical pathology (37 versus 22 mm) and axillary lymph node positivity (41.3 versus 22.8 percent) were features associated with interval cancers detected by AI. AI correctly localized 84.4 percent of screening-detected cancers using the same threshold (score ≥10), and correctly classified 85.9 and 73.3 percent of true-negative and false-positive cases as negative.

“Our study shows that an AI algorithm can retrospectively detect and correctly localize nearly one-third of interval breast cancers on screening DBT exams, suggesting its potential to reduce the interval cancer rate and improve screening outcomes,” Bahl said in a statement.

One author disclosed ties to industry; the study was funded by Lunit.

Abstract/Full Text

Editorial (subscription or payment may be required)