← Accueil actu | Salle du marché
2026-09-10 · Source：Unite AI
Researchers at NYU Langone Health and its Perlmutter Cancer Center unveiled NYU-DRP, a deep-learning tool that estimates a woman's five-year breast cancer risk from longitudinal 3D mammograms.
NYU Langone Health announced on September 10, 2026, that researchers at the institution and its Perlmutter Cancer Center have built NYU-DRP, a deep-learning tool that estimates a woman's five-year risk of developing breast cancer. Where risk assessment might otherwise hinge on a single examination, the system draws its predictions from 3D mammograms captured across multiple years, treating a patient's imaging history itself as the predictive raw material.
The tool's technical foundation is longitudinal digital breast tomosynthesis, known as longitudinal DBT — the 3D mammogram record a woman accumulates over successive screening visits. Rather than evaluating an isolated scan, the model reads this multi-year archive, positioning the accumulation of imaging data over time as central to risk estimation. The distinguishing feature of the work, as described in the announcement, is precisely this longitudinal strategy: the model is designed to learn from how a patient's 3D mammography record extends across years, not from any one frame within it.
For readers tracking AI compute demand, the announcement underscores that medical imaging remains one of the most active fronts for deep-learning deployment. Models of this kind depend on training against large, longitudinal image archives — a workload profile distinct from the single-prompt, single-inference patterns that dominate consumer-facing AI. Hospital-embedded research groups working alongside dedicated cancer centers, as at NYU Langone and its Perlmutter Cancer Center, are positioned as notable producers of clinically aimed models.
The core claim is narrow but concrete: NYU-DRP converts years of accumulated 3D mammography into a five-year breast cancer risk estimate. By anchoring prediction in longitudinal DBT, the NYU team frames risk assessment around the patient's imaging record as a continuous history rather than a series of disconnected exams — an approach that places multi-year image archives, and the compute needed to learn from them, at the center of the model's design.
Mots-clés: NYU Langone Health · NYU-DRP · Perlmutter Cancer Center · Deep Learning · Medical Imaging · Breast Cancer
萬安算交所 AIXX · AIXX · Bourse compute vérifiée · Accéder à la salle