A Fully Open and Generalizable Foundation Model for Ultrasound Clinical Applications

2025-09-15

AI Summary

Poor generalizability of clinical ultrasound AI models stems primarily from scarce high-quality annotated data and the task-specific nature of existing architectures. To address this, we propose the first open, general-purpose foundation model tailored for ultrasound imaging. It is pretrained via self-supervised learning on EchoCareData—a large-scale, publicly available dataset comprising multi-center, multi-device, and multi-ethnic ultrasound scans. Our model introduces a novel hierarchical classifier architecture that jointly captures pixel-level local features and representation-level global anatomical structures. The framework supports both downstream fine-tuning and localized adaptation. Evaluated across ten benchmark tasks—including diagnosis, segmentation, detection, quantitative analysis, and report generation—it consistently surpasses state-of-the-art methods, demonstrating plug-and-play, high-performance cross-scenario transferability.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Diffusion Models for Vision
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Artificial intelligence (AI) that can effectively learn ultrasound representations by integrating multi-source data holds significant promise for advancing clinical care. However, the scarcity of large labeled datasets in real-world clinical environments and the limited generalizability of task-specific models have hindered the development of generalizable clinical AI models for ultrasound applications. In this study, we present EchoCare, a novel ultrasound foundation model for generalist clinical use, developed via self-supervised learning on our curated, publicly available, large-scale dataset EchoCareData. EchoCareData comprises 4.5 million ultrasound images, sourced from over 23 countries across 5 continents and acquired via a diverse range of distinct imaging devices, thus encompassing global cohorts that are multi-center, multi-device, and multi-ethnic. Unlike prior studies that adopt off-the-shelf vision foundation model architectures, we introduce a hierarchical classifier into EchoCare to enable joint learning of pixel-level and representation-level features, capturing both global anatomical contexts and local ultrasound characteristics. With minimal training, EchoCare outperforms state-of-the-art comparison models across 10 representative ultrasound benchmarks of varying diagnostic difficulties, spanning disease diagnosis, lesion segmentation, organ detection, landmark prediction, quantitative regression, imaging enhancement and report generation. The code and pretrained model are publicly released, rendering EchoCare accessible for fine-tuning and local adaptation, supporting extensibility to additional applications. EchoCare provides a fully open and generalizable foundation model to boost the development of AI technologies for diverse clinical ultrasound applications.

Problem

• Develops a generalizable AI foundation model for ultrasound clinical applications
• Addresses scarcity of large labeled datasets and limited model generalizability
• Enables diverse tasks like diagnosis, segmentation, and report generation

Innovation

• Self-supervised learning on large-scale ultrasound dataset
• Hierarchical classifier for joint feature learning
• Open foundation model for diverse clinical applications