The Food and Drug Administration's Center for Devices and RadiologicalHealth (CDRH) has established a new working group to deal withregulatory issues related to computer-aided diagnosis (CAD) software.In announcing the move, the agency said it has seen
The Food and Drug Administration's Center for Devices and RadiologicalHealth (CDRH) has established a new working group to deal withregulatory issues related to computer-aided diagnosis (CAD) software.In announcing the move, the agency said it has seen an increasingnumber of regulatory applications for CAD software products andis interested in soliciting comment from manufacturers as to howthese products should be regulated.
CAD software uses image processing algorithms to detect suspiciousareas on digitized images and to highlight those areas for theattention of interpreting physicians. Mammography is one areato which CAD techniques are being applied, and companies suchas Stereometrix, R2 Technology and MedDetect are developing CADmammography software.
The FDA is trying to determine what regulatory path to takewith CAD applications, such as whether they should go throughthe premarket approval (PMA) process or the 510(k) notificationprocess, according to David Brown, chief scientist of the divisionof electronics and computer science in CDRH.
"We'd like to be specific and give good guidance to reviewersabout what claims need to be backed up by what evidence, and whatclaims need to be backed up by the PMA process," Brown said.
A meeting on the subject has been tentatively scheduled forDec. 19 at the FDA's Parklawn Building at 5600 Fishers Lane inRockville, MD. Vendors and clinicians interested in providinginput to the CAD working group should contact either David Brownby e-mail at dgb@fdadr.cdrh.fda.gov, or Mary Anderson at mpa@fdadr.cdrh.fda.gov.Comments can be faxed to 301/443-9101.
Considering Breast- and Lesion-Level Assessments with Mammography AI: What New Research Reveals
June 27th 2025While there was a decline of AUC for mammography AI software from breast-level assessments to lesion-level evaluation, the authors of a new study, involving 1,200 women, found that AI offered over a seven percent higher AUC for lesion-level interpretation in comparison to unassisted expert readers.