UPCMR: A Universal Prompt-guided Model for Random Sampling Cardiac MRI Reconstruction

2025-02-18

AI Summary

Cardiac magnetic resonance imaging (CMR) offers high diagnostic value but suffers from long acquisition times; acceleration via k-space undersampling often degrades reconstruction quality. Existing deep learning methods exhibit poor generalization across diverse sampling patterns and acceleration factors. To address this, we propose an end-to-end universal reconstruction framework featuring a novel dual-prompt mechanism—undersampling-specific and spatial-specific prompts—embedded into every layer of a U-Net backbone, enabling single-model adaptability to multiple sampling configurations. The framework integrates a learnable prompt-based unrolled network with k-space data consistency constraints and is trained on the CMRxRecon2024 dataset. In comprehensive random undersampling scenarios, our method achieves significantly higher PSNR and SSIM than both conventional and state-of-the-art deep learning approaches, demonstrating strong generalization capability and promising clinical deployability.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Cardiac magnetic resonance imaging (CMR) is vital for diagnosing heart diseases, but long scan time remains a major drawback. To address this, accelerated imaging techniques have been introduced by undersampling k-space, which reduces the quality of the resulting images. Recent deep learning advancements aim to speed up scanning while preserving quality, but adapting to various sampling modes and undersampling factors remains challenging. Therefore, building a universal model is a promising direction. In this work, we introduce UPCMR, a universal unrolled model designed for CMR reconstruction. This model incorporates two kinds of learnable prompts, undersampling-specific prompt and spatial-specific prompt, and integrates them with a UNet structure in each block. Overall, by using the CMRxRecon2024 challenge dataset for training and validation, the UPCMR model highly enhances reconstructed image quality across all random sampling scenarios through an effective training strategy compared to some traditional methods, demonstrating strong adaptability potential for this task.

Problem

• Accelerates cardiac MRI scanning
• Improves image reconstruction quality
• Adapts to various sampling modes

Innovation

• Universal unrolled model
• Learnable prompts integration
• UNet structure enhancement