Overview
- At a CADIME seminar on Monday, presenters reported that AI reduced by up to 50% the time radiologists spend reviewing mammograms that show no suspicious findings.
- The tools have been integrated into workflows at 77 public hospitals and have run 21,042 unique historical analyses as part of national PACS modernization programs.
- Hospitals and developers described main use cases as mass oncologic screening, automated clinical reporting, rapid triage in emergency services, and a ‘second opinion’ mode to flag areas for closer review.
- Developers said systems were trained on more than one million images and can reach sensitivity above 95% in some modes, but project leads and experts stressed that independent validation and continuous operational monitoring remain needed.
- Project leaders reported large reported cost savings under the pay‑per‑study model and said faster triage should let radiologists focus on higher‑risk cases, which could reduce missed cancers if clinical oversight and ongoing optimization continue.