VinDr-Mammo
VinBigData Joint Stock CompanyVinDr-Mammo is a software tool designed to assist radiologists in prioritizing mammogram exams by flagging those that may contain suspicious findings. It analyzes 2D digital mammograms using artificial intelligence to identify potentially concerning cases and alerts radiologists, enabling more efficient workflow and faster patient care. It does not replace the physician's final diagnosis.
Intended Use
The VinDr-Mammo is a passive notification for prioritization-only, a parallel-workflow software tool used by MQSA qualified interpreting physicians to prioritize patients with suspicious findings in the medical care environment. It analyzes 2D FFDM screening mammograms and alerts cases suggestive of suspicious findings to a PACS/Workstation to allow worklist prioritization. It does not provide diagnostic information or remove images from the physician’s worklist and is intended only for use with validated FFDM systems.
Technology
VinDr-Mammo is a software-as-a-medical-device (SaMD) that employs a machine learning algorithm trained on diverse datasets to analyze 2D FFDM mammograms and detect exam-level suspicious findings. The software accepts DICOM studies, validates input data (e.g., patient age, modality, view), and produces exam-level notification codes delivered via an API to PACS or workstations for worklist prioritization in radiology workflows. It operates as a passive notification tool and does not alter the standard of care workflow or remove cases from the worklist.
Performance
The performance of VinDr-Mammo was validated using two datasets: a US population dataset from RSNA with 1000 cases (252 cancer positive) and a Vietnamese hospital dataset with 1864 cases (466 cancer positive). The device demonstrated high sensitivity (~90%), specificity (~91%), and AUC (~0.96), consistent across breast density, age groups, scanner models, and lesion types. The device processes mammograms and returns results with an average time of 2.8 minutes, meeting clinical workflow needs. Performance was shown to be substantially equivalent to its predicate device (CogNet QmTRIAGE, K220080).
Predicate Devices
No predicate devices specified
Device Timeline
- 1
Submission
9/27/2023
7 months - 2
FDA Approval
5/23/2024
Ready to Sharpen Your Edge?
Join hundreds of your peers who rely on RadAI Slice. Get the essential weekly briefing that empowers you to navigate the future of radiology.
We respect your privacy. Unsubscribe at any time.