BreastNegative – The AI That Accelerates and Optimizes Mammographic Screening
BreastNegative is the solution developed by Health Triage to optimize breast cancer screening programs. It uses neural networks based on deep learning models trained on millions of real images and validated by independent scientific institutions.
The system was born from the collaboration between Health Triage and researchers from the Department of Automatic Control and Computer Engineering at Politecnico di Torino. The goal is to improve the quality of oncological prevention through reliable, certified technology that integrates seamlessly into clinical workflows.
A concrete answer to the limitations of traditional screening
Over 99% of screening mammograms turn out to be negative. A reassuring figure for women, but one that leads to inefficient use of medical resources — in a system already marked by a shortage of radiologists and significant regional disparities in access to prevention programs.
This is where BreastNegative comes in: it automatically analyzes every mammogram and identifies with high accuracy those exams showing no suspicious findings. Mammograms classified as definitively negative, according to validated clinical criteria, are reviewed by a single radiologist instead of two, freeing up resources that can be redirected toward expanding existing screening programs and/or managing particularly complex cases.
What sets it apart from other AI systems
Not just detecting tumors, but safely ruling out healthy cases
Unlike most AI solutions for mammography, which focus on identifying lesions, BreastNegative was designed to reliably recognize negative cases.
This approach makes it possible to:
- increase the operational efficiency of screening centers;
- optimize the use of radiological resources;
- improve coverage of prevention programs.
BreastNegative: support for breast diagnostics with intelligent triage and negativity assessment
BreastNegative, an AI-based software medical device, is intended for use by medical professionals — specifically radiologists specialized in breast imaging — to support the clinical evaluation of mammographic exams.
The device takes a complete mammographic exam as input and, by analyzing the radiological images, provides a negativity classification output that categorizes the exam result into three classes:
- Negative: the analyzed exam shows no signs of breast cancer.
- Double reading needed: the device cannot determine whether a tumor is present or absent.
- Not processable: the device was unable to process the exam.
In a mammographic screening context based on double-blind reading, the “Negative” output is designed to support clinical workflow optimization by reducing the need for a second reading by breast radiologists. When the output is “Double reading needed”, the exam is referred back to standard clinical practice.
⚠️ The software does not formulate diagnoses autonomously, nor does it replace clinical judgment. The responsibility for the diagnostic decision and the final report remains exclusively with the specialist physician.


