A Versatile Foundation Model for AI-enabled Mammogram Interpretation

2025-09-24

AI Summary

Current breast X-ray foundation models suffer from insufficient data diversity, poor generalizability, and inadequate clinical evaluation. To address these limitations, we propose a two-stage pretraining paradigm integrating self-supervised learning with clinical knowledge distillation, yielding the first general-purpose foundation model specifically designed for mammographic imaging. Trained on a large-scale, multicenter dataset, the model supports diverse downstream tasks—including lesion detection, segmentation, classification, image retrieval, and visual question answering—within a unified framework. We further introduce a comprehensive benchmark comprising 92 clinically relevant tasks. Extensive experiments demonstrate superior performance: the model achieves first place on 50 of 68 internal validation tasks (mean rank = 1.5) and on 20 of 24 external validation tasks (mean rank = 1.2), significantly outperforming existing methods. This work advances automated, clinically translatable early breast cancer screening.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Other Foundations of Machine Learning
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web search models and ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer-related mortality in women globally. Mammography is essential for the early detection and diagnosis of breast lesions. Despite recent progress in foundation models (FMs) for mammogram analysis, their clinical translation remains constrained by several fundamental limitations, including insufficient diversity in training data, limited model generalizability, and a lack of comprehensive evaluation across clinically relevant tasks. Here, we introduce VersaMammo, a versatile foundation model for mammograms, designed to overcome these limitations. We curated the largest multi-institutional mammogram dataset to date, comprising 706,239 images from 21 sources. To improve generalization, we propose a two-stage pre-training strategy to develop VersaMammo, a mammogram foundation model. First, a teacher model is trained via self-supervised learning to extract transferable features from unlabeled mammograms. Then, supervised learning combined with knowledge distillation transfers both features and clinical knowledge into VersaMammo. To ensure a comprehensive evaluation, we established a benchmark comprising 92 specific tasks, including 68 internal tasks and 24 external validation tasks, spanning 5 major clinical task categories: lesion detection, segmentation, classification, image retrieval, and visual question answering. VersaMammo achieves state-of-the-art performance, ranking first in 50 out of 68 specific internal tasks and 20 out of 24 external validation tasks, with average ranks of 1.5 and 1.2, respectively. These results demonstrate its superior generalization and clinical utility, offering a substantial advancement toward reliable and scalable breast cancer screening and diagnosis.

Problem

• Developing a versatile foundation model for mammogram interpretation to overcome limitations in existing AI systems
• Addressing insufficient training data diversity and limited model generalizability in mammogram analysis
• Establishing comprehensive evaluation across multiple clinically relevant breast cancer detection tasks

Innovation

• Two-stage pre-training with self-supervised and supervised learning
• Knowledge distillation to transfer features and clinical knowledge
• Comprehensive benchmark with 92 tasks across five clinical categories