HIVE Mammography Image Processing Demo
Research Prototype · Automated Mammography Analysis
Published Research Behind This Demo
This demo is part of a research effort led by Asma Baccouche, Bilal Ghalib-Zouari, Yassine Zemmouri and others, published in Computer Methods and Programs in Biomedicine (Elsevier, 2022). The paper investigates an end-to-end YOLO-based fusion framework to simultaneously detect and classify suspicious findings on digital mammograms, including Mass, Calcification, Architectural Distortion, and Normal cases.
In addition to analysis of current mammograms, the study explores retrospective early prediction by generating synthetic prior mammograms using image-to-image translation techniques (CycleGAN and Pix2Pix). The trained detection model is then applied to these translated prior images to investigate whether subtle patterns associated with future abnormalities can be identified before clinical diagnosis.
For demonstration purposes only. Not for clinical use or medical decision-making.
What Happens Behind the Scenes
1 — Locate
The system scans the mammogram and identifies regions that may contain suspected abnormalities such as masses or calcifications.
2 — Focus
When a suspicious region is found, it is isolated and outlined, making the area easier to visualize independently of surrounding breast tissue.
3 — Analyze
The isolated region is processed by a deep-learning model that generates an automated, image-based assessment to illustrate how AI can assist radiologists in future diagnostic workflows.
Try the Demo in 4 Steps
- Step 1: Enter the email address where you want to receive the result.
- Step 2: Select a mammography image from your device (e.g., JPG, PNG, or another supported format).
- Step 3: Click “Upload & Send Result” to send the image to the processing server.
- Step 4: Wait a few moments, then check your email for the processed result. If it does not appear, check your spam or junk folder.
This system is a research demonstration prototype. It is not FDA-approved and is not intended to diagnose, treat, or monitor any disease. Results may be incomplete or inaccurate and must not replace clinical judgment.