Bunkerhill AVC

Bunkerhill AVC is a software device that automatically detects and quantifies aortic valve calcification in non-gated, non-contrast chest CT images for patients aged 40 and above. It outputs an estimated Agatston-equivalent calcium score and a circular region of interest showing the location of detected calcium, intended as adjunctive information to assist physicians during case review without replacing the original CT scan or clinical judgment.

Both the subject device and predicate (iCAC) use deep-learning algorithms to analyze DICOM-formatted non-gated chest CT images and generate Agatston-equivalent scores. Key differences: Bunkerhill AVC detects aortic valve calcium and provides a binary presence/absence output with a circular ROI, whereas iCAC detects coronary artery calcium and provides segmentation with four risk categories. Both support slice thickness up to 5 mm and optional reporting to patient records.

Not stated in this summary. The document references FDA guidance on software submissions ('Content of Premarket Submissions for Device Software Functions') but does not cite specific ISO, IEC, or ASTM consensus standards.

Both devices use identical technological approaches (deep learning, DICOM input, Agatston-equivalent scoring, adjunctive physician workflow integration) on the same imaging modality and anatomy (chest CT). The anatomical difference—aortic valve versus coronary arteries—does not raise new safety or effectiveness concerns because both perform the same functional task: detecting and localizing calcification to assist clinical review. Performance testing demonstrated Bunkerhill AVC achieved acceptance criteria for mean difference and limits of agreement (bias −5.15 AU, LoA −200.96 to 190.65 AU) comparable to predicate device performance, satisfying the primary endpoint for substantial equivalence under 21 CFR 807.92(b)(3).