Saige-Dx

Saige-Dx is a computer-assisted detection and diagnosis software designed to enhance the identification of suspicious breast lesions and calcifications on digital breast tomosynthesis mammograms. Utilizing image processing and AI-driven algorithms, the device assigns suspicion levels to findings, validated through comprehensive testing to demonstrate safety and effectiveness for screening mammography in women aged 35 and older.

Saige-Dx, a radiological computer-assisted detection and diagnosis (CAD/CDD) software developed by DeepHealth, Inc., has received regulatory clearance, affirming its substantial equivalence to existing devices. This permits its market entry as a concurrent reading aid for interpreting physicians analyzing digital breast tomosynthesis (DBT) mammograms, specifically designed to identify soft tissue lesions and calcifications indicative of potential cancer. Its primary purpose is to augment diagnostic accuracy in screening mammography for women aged 35 and older. The system functions by processing DBT image stacks and accompanying 2D images, subsequently assigning a "Suspicion Level" for both individual findings and entire cases. Recent advancements enable unilateral image analysis, examination of breast implant patients via implant displaced views, and compatibility with various vendor equipment like Hologic HD and GE. Extensive verification and standalone performance validation on a robust and diverse dataset, adhering to recognized standards, confirmed Saige-Dx's safety, effectiveness, and comparable performance to predicate devices, supporting its intended clinical use.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What are the specific sensitivity and specificity values demonstrated by Saige-Dx in its standalone performance validation, particularly for different lesion types (masses vs. calcifications) and breast densities?; How does the "Suspicion Level" generated by Saige-Dx correlate with the positive predictive value of biopsies in a clinical setting?; Have studies been conducted to evaluate the impact of Saige-Dx on radiologist reading time, confidence, and overall diagnostic accuracy when used as a concurrent reading aid?; What is the clinical utility and accuracy of Saige-Dx when analyzing unilateral images or images from patients with breast implants, given these are recent updates?; What post-market surveillance or real-world evidence will be collected to monitor Saige-Dx's performance and identify any potential biases or limitations in diverse patient populations or clinical workflows?