CogNet AI-MT+

CogNet AI-MT+ is cloud-based software that uses artificial intelligence to analyze Digital Breast Tomosynthesis (DBT) screening mammograms and flags studies with suspicious findings for prioritization. It is a passive notification tool for MQSA-qualified interpreting physicians that aids in worklist prioritization and does not replace physician evaluation or make diagnostic determinations.

The main difference from the predicate device (CogNet AI-MT) is that CogNet AI-MT+ processes DBT images with multiple slices rather than 2D FFDM flat images. Both devices use similar AI algorithms, operate in parallel workflow, produce passive notifications to PACS/Workstation, are cloud-based, and exclude patients with breast implants and post-surgical studies. The predicate outputs an encapsulated PDF while the new device outputs a DICOM secondary capture image.

IEC 62304 (Software Development Life Cycle), 21 CFR Part 820 (FDA Regulatory Requirements and Design Controls), and DICOM PS3.1 (Digital Image and Communications in Medicine).

CogNet AI-MT+ is substantially equivalent to the predicate because both devices share the same regulatory classification, intended use (passive notification for prioritization), user (MQSA physicians), product code, and core AI functionality. Performance testing showed the new device achieved AUROC of 0.9548 with sensitivity 0.8809 and specificity 0.9156, meeting or exceeding the same objectives as the predicate (AUROC 0.9569). The difference in image modality (DBT vs. FFDM) does not raise new safety or effectiveness concerns.