BunkerHill BMD

Bunkerhill BMD is a software-only medical device using deep-learning algorithms to estimate bone mineral density from existing CT scan images. It is intended for adults 30 years and above to assess spinal bone density and flag low bone density below a pre-specified threshold, supporting opportunistic retrospective bone density assessment without requiring dedicated imaging or a phantom.

Both the subject and predicate devices use deep-learning algorithms to estimate average bone mineral density from CT images in DICOM format, serve as clinical decision-support tools without replacing clinical evaluation, and generate calibrated BMD reports with T-scores and Z-scores. The subject device provides a subset of predicate outputs (T-score group only) and automatically measures Hounsfield units across spinal bone regions of interest. Both have identical product code (KGI), regulation number (21 CFR 892.1170), and modality (computed tomography).

Not stated in this summary.

Bunkerhill BMD achieves substantial equivalence because it shares the same intended use, indications, technological principles, and regulatory classification as the predicate ABMD device. Both use identical deep-learning methodology for spinal bone density estimation from DICOM CT images and serve as physician support tools rather than replacements for clinical judgment. The subject device's validation study demonstrates sensitivity of 81.0% and specificity of 78.4% across 371 CT studies from geographically diverse sites with consistent performance across patient subgroups (age, sex, CT manufacturer, slice thickness, reconstruction kernel), and the minor differences in output scope do not introduce new safety or effectiveness concerns.