Manisha Bahl, Saul Langarica, Leslie R. Lamb, Ariel S. Kniss, Synho Do<\/p>\n
Radiology, 2025<\/strong><\/p>\n Abstract<\/strong> Background<\/strong> Purpose<\/strong> Materials and Methods<\/strong>
\nThe use of a U.S. Food and Drug Administration\u2013cleared artificial intelligence algorithm for lesion detection and diagnosis at screening digital breast tomosynthesis examinations could reduce interval cancers by nearly one-third.<\/p>\n
\nGiven the lack of long-term outcome data for screening digital breast tomosynthesis (DBT), the interval cancer rate is a commonly used surrogate for patient outcomes.<\/p>\n
\nTo evaluate the performance of an artificial intelligence (AI) algorithm in detecting and localizing interval cancers at screening DBT and to validate the diagnostic threshold of the algorithm by analyzing its performance across interval, true-positive, true-negative, and false-positive screening DBT examinations.<\/p>\n
\nScreening DBT examinations performed immediately before confirmed interval cancer diagnoses between February 2011 and June 2023 at an academic institution were retrospectively analyzed by a U.S. Food and Drug Administration\u2013cleared AI algorithm. Lesions marked by AI on DBT sections were assigned a score from 0 to 100, with the examination-level score reflecting the highest lesion score. AI-positive examinations (score \u226510) were independently reviewed by two breast imaging radiologists to determine whether AI annotations corresponded to the site of the subsequent cancer. Imaging and clinicopathologic features were compared between interval cancers detected and not detected by AI using the Wilcoxon signed rank test and Fisher exact test. Using the same threshold score, 1000 true-positive, true-negative, and false-positive screening DBT examinations were also analyzed by AI.<\/p>\n