Saige-Dx
Overview
DeepHealth’s Saige-Dx is an advanced AI-powered mammography diagnostic software designed to assist radiologists in the more effective detection of breast cancer. Built on sophisticated deep-learning algorithms, Saige-Dx automatically identifies suspicious lesions in mammograms and assigns a suspicion level to both individual findings and the entire case. This technology aims to enhance diagnostic accuracy, enable earlier cancer detection, and reduce unnecessary patient recalls. It analyzes both digital breast tomosynthesis (DBT) and 2D mammograms, providing critical insights to interpreting physicians. Saige-Dx is a core component of RadNet’s Enhanced Breast Cancer Detection (EBCD) service, demonstrating improved radiologist performance in multi-reader studies. The software is designed to integrate into existing radiology workflows, offering capabilities like case prioritization for high-suspicion exams.
Reviewed by Pouyan Golshani, MD — Interventional Radiologist
Key Features
- AI-powered detection of suspicious lesions in mammograms
- Assigns a suspicion level to each finding and the entire case
- Helps detect and diagnose breast cancer earlier
- Reduces unnecessary recalls
- Improves radiologist performance (increased cancer detection, lower false positive rate)
- Analyzes digital breast tomosynthesis (DBT) and 2D mammograms
- Outputs bounding boxes circumscribing detected findings
- Provides case-level and finding-level outputs
- Compatible with various DBT hardware
- Can be used for triage (prioritizing suspicious cases)
Use Cases
- Breast cancer screening and early detection
- Assisting radiologists in mammography interpretation
- Improving diagnostic accuracy and efficiency in breast imaging
- Reducing false positives and unnecessary recalls in mammography
- Prioritizing high-suspicion mammograms for faster review
- Integration into Enhanced Breast Cancer Detection (EBCD) services
What Physicians Need to Know
Saige-Dx is an adjunct tool and should not replace a physician's own review of a mammogram; decisions should not be made solely based on its analysis. While it can improve reader performance and potentially reduce reading time, the exact amount of time saved in real-world use has not been prospectively studied. The ability to process prior exams can help reduce false positives, which is a significant issue in radiology. Consider its use as a concurrent reading aid to augment your interpretation and potentially detect subtle lesions earlier.
Saige-Dx integrates into existing clinical workflows by generating DICOM Structured Reports and Secondary Capture objects that can be overlaid on original mammogram images using standard viewing workstations. It is compatible with various mammography systems, including those from GE HealthCare and Siemens.
Details
| Category | Oncology AI, Radiology & Imaging AI |
| Pricing | Contact for pricing |
| Deployment | Cloud-native, on-premise, or hybrid |
| Compliance | |
| BAA Available | Yes AI-estimated |
| HIPAA Compliant | Yes AI-estimated |
| FDA Status | Cleared AI-estimated |
| Integrations | |
| EHR | Not specified |
| Specialties | Oncology, Radiology |
What the Web Says
Saige-Dx by DeepHealth is an FDA-cleared AI-powered software designed to assist radiologists in detecting breast cancer by analyzing mammograms, including both digital breast tomosynthesis (DBT) and 2D images. It identifies suspicious lesions, assigns suspicion levels for findings and cases, and is intended to be used as a concurrent reading aid for interpreting physicians. The software has been validated through multiple studies, demonstrating improved performance in detecting breast cancer and is integrated into RadNet's Enhanced Breast Cancer Detection (EBCD) service.
Overall: PositiveStrengths
- Improved diagnostic accuracy and efficiency for radiologists in detecting breast cancer.
- Reduces false alarms and helps in finding early cancers.
- Enhances decision-making processes with AI-driven insights.
- Seamless integration with existing systems and intuitive user interface.
- Trained on a vast dataset, including over a million images and thousands of biopsy-proven cancers, from diverse populations.
- Can lower recall rates while increasing cancer detection rates.
Limitations
- It is an adjunct tool and not intended to replace a physician's own review of a mammogram; decisions should not be made solely based on AI analysis.
- Independent proof of benefit is considered thin, as the main reader study was conducted by the vendor's own employees.
- The case set for studies was retrospective and enriched, meaning it measures reader assistance on a curated sample rather than real-world screening outcomes.
- Some services utilizing Saige-Dx, like RadNet's EBCD program, may incur an additional fee for patients.
- While increasing cancer detection, a multi-stage AI-driven workflow with Saige-Dx also led to a 5.7% increased recall rate.
- The portfolio is still being merged, and groups desiring independent proof or a single product rather than a roadmap might need to wait.
Based on reviews from: fda.gov, myebcdmammo.com, appliedradiology.com, computerreview.com, diagnosticimaging.com, healthaidb.com, medaiverdict.com, vercel.app, auntminnie.com, nih.gov, fiercebiotech.com, radnet.com, stocktitan.net, researchgate.net
Last updated: 2026-10-29
Ratings & Reviews
No reviews yet. Be the first to review this tool!
Rate Saige-Dx
Press & Coverage
Videos
Product demos, reviews, and walkthroughs for Saige-Dx.
Why You Should Include Artificial Intelligence with Your Next Mammogram | Radiology Imaging Assoc
Frequently Asked Questions
Investors who backed Saige-Dx
Funded through the company that built this tool.
