Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications

2025-08-09

AI Summary

Deep learning models for multi-sequence MRI suffer from poor cross-protocol generalization due to acquisition parameter heterogeneity. Method: We propose PRISM, a pretraining paradigm that employs self-supervised learning to disentangle anatomy-invariant features from sequence-specific variations, accompanied by the largest publicly reported multi-organ, multi-sequence MRI pretraining dataset. Our approach integrates large-scale self-supervised pretraining, feature disentanglement, and high-dimensional semantic preservation. Contribution/Results: We systematically evaluate PRISM on 44 downstream tasks—including segmentation, registration, diagnosis, and report generation—across 32 public datasets and 5 private cohorts. PRISM achieves state-of-the-art performance on 39 tasks, significantly outperforming existing foundation models and non-pretrained baselines. It establishes a robust, scalable, and generalizable representation foundation for multi-sequence MRI analysis.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Multi-sequence Magnetic Resonance Imaging (MRI) offers remarkable versatility, enabling the distinct visualization of different tissue types. Nevertheless, the inherent heterogeneity among MRI sequences poses significant challenges to the generalization capability of deep learning models. These challenges undermine model performance when faced with varying acquisition parameters, thereby severely restricting their clinical utility. In this study, we present PRISM, a foundation model PRe-trained with large-scale multI-Sequence MRI. We collected a total of 64 datasets from both public and private sources, encompassing a wide range of whole-body anatomical structures, with scans spanning diverse MRI sequences. Among them, 336,476 volumetric MRI scans from 34 datasets (8 public and 26 private) were curated to construct the largest multi-organ multi-sequence MRI pretraining corpus to date. We propose a novel pretraining paradigm that disentangles anatomically invariant features from sequence-specific variations in MRI, while preserving high-level semantic representations. We established a benchmark comprising 44 downstream tasks, including disease diagnosis, image segmentation, registration, progression prediction, and report generation. These tasks were evaluated on 32 public datasets and 5 private cohorts. PRISM consistently outperformed both non-pretrained models and existing foundation models, achieving first-rank results in 39 out of 44 downstream benchmarks with statistical significance improvements. These results underscore its ability to learn robust and generalizable representations across unseen data acquired under diverse MRI protocols. PRISM provides a scalable framework for multi-sequence MRI analysis, thereby enhancing the translational potential of AI in radiology. It delivers consistent performance across diverse imaging protocols, reinforcing its clinical applicability.

Problem

• Improves generalization of deep learning models for multi-sequence MRI analysis
• Addresses heterogeneity in MRI sequences affecting model performance
• Enhances clinical utility of AI across diverse MRI protocols

Innovation

• Large-scale multi-sequence MRI pretraining corpus
• Disentangles invariant and sequence-specific features
• Scalable framework for diverse clinical tasks