MammoScreen BD

MammoScreen BD is a software application that uses artificial intelligence to automatically assess breast density in mammograms and digital breast tomosynthesis images. It provides an ACR BI-RADS 5th Edition breast density category to aid radiologists interpreting screening mammograms in asymptomatic women age 40 and older, functioning as adjunctive information only and not as a diagnostic tool.

The subject device expands mammogram system compatibility compared to the predicate by supporting GE and Siemens mammograms in addition to Hologic systems. Both devices use supervised machine learning with deep learning modules trained on large databases of annotated mammograms. The subject device includes a Predetermined Change Control Plan (PCCP) for future modifications including additional manufacturer support and unsupervised pre-training of the neural network backbone.

IEC 62304:2006/A1:2016 (Medical device software — Software life-cycle processes) and IEC 62366-1:2015+AMD1:2020 (Medical devices — Application of usability engineering to medical devices).

The subject device maintains identical indications for use, intended populations, anatomical location, and fundamental AI architecture as the predicate. Both assess breast density via supervised machine learning and output ACR BI-RADS categories at the mammogram level. The subject device's expanded compatibility with additional mammogram manufacturers (GE, Siemens) does not raise different safety or effectiveness questions because performance testing demonstrates non-inferior or superior quadratically weighted Cohen's kappa values (ranging from 89.03 to 93.19) across all manufacturers, exceeding the pre-determined reference threshold of 0.85, with consistent performance across patient age groups, races, and breast thicknesses.