Saige-Dx

Saige-Dx is artificial intelligence software that analyzes digital breast tomosynthesis (DBT) mammograms to detect soft tissue lesions and calcifications potentially indicative of cancer. It assigns suspicion levels to individual findings and an overall case suspicion level, providing results as a concurrent reading aid for qualified radiologists interpreting screening mammograms. The device is not intended to replace physician judgment.

Compared to the predicate device (Saige-Dx v.3.1.0), the subject device features an updated AI algorithm, an added output configuration, improved acceptance checks, refinements to improve usability, and improved memory and GPU management. Both devices use machine learning to analyze mammography images and output bounding boxes with finding suspicion levels and case-level assessments in DICOM Structured Report format.

ISO 14971:2019 (Risk Management), IEC 62304:2015 (Software Life Cycle Processes), NEMA PS3 (DICOM), FDA Guidance on Premarket Submissions for Software in Medical Devices (May 2005), and FDA Guidance on Software as a Medical Device Clinical Evaluation (December 2017).

The subject and predicate devices have identical indications for use, intended user populations, and intended patient populations. Verification and validation testing on 2,002 DBT screening mammograms demonstrated non-inferior performance compared to the predicate device, with the algorithm meeting pre-specified performance criteria across diverse breast densities, patient demographics, and lesion types. The differences (updated algorithm, output configuration, usability refinements) do not alter the intended use or clinical performance and therefore do not affect safety and effectiveness.