A Unified Model for Compressed Sensing MRI Across Undersampling Patterns

2024-10-05

AI Summary

Conventional compressed sensing MRI (CS-MRI) reconstruction models require retraining for each undersampling pattern and image resolution, hindering clinical flexibility and deployment. Method: We propose the first generalizable unified reconstruction model, featuring a novel neural-operator-based dual-space (image domain + measurement domain) architecture that is inherently discretization-agnostic—enabling zero-shot support for arbitrary undersampling masks, super-resolution, and field-of-view extension without dependence on specific sampling patterns or resolutions. Contribution/Results: Compared to end-to-end VarNet, our method achieves average improvements of 11% in SSIM and 4 dB in PSNR; it is 1,400× faster than diffusion-based methods at inference. It demonstrates strong robustness across varying acceleration factors and resolutions. This work overcomes the customization bottleneck of traditional CNN-based CS-MRI models and establishes a new paradigm for real-time, adaptive clinical MRI acceleration.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Learning on the Edge & Model Compression
• Cognitive Modeling & Cognitive Systems: (Computational) Cognitive Architectures

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Compressed Sensing MRI reconstructs images of the body's internal anatomy from undersampled measurements, thereby reducing the scan time - the time subjects need to remain still. Recently, deep neural networks have shown great potential for reconstructing high-fidelity images from highly undersampled measurements in the frequency space. However, one needs to train multiple models for different undersampling patterns and desired output image resolutions, since most networks operate on a fixed discretization. Such approaches are highly impractical in clinical settings, where undersampling patterns and image resolutions are frequently changed to accommodate different real-time imaging and diagnostic requirements. We propose a unified model robust to different measurement undersampling patterns and image resolutions in compressed sensing MRI. Our model is based on neural operators, a discretization-agnostic architecture. Neural operators are employed in both image and measurement space, which capture local and global image features for MRI reconstruction. Empirically, we achieve consistent performance across different undersampling rates and patterns, with an average 11 percent SSIM and 4dB PSNR improvement over a state-of-the-art CNN, End-to-End VarNet. For efficiency, our inference speed is also 1,400x faster than diffusion methods. The resolution-agnostic design also enhances zero-shot super-resolution and extended field of view in reconstructed images. Our unified model offers a versatile solution for MRI, adapting seamlessly to various measurement undersampling and imaging resolutions, making it highly effective for flexible and reliable clinical imaging. Our code is available at https://armeet.ca/nomri.

Problem

• Develops a unified MRI reconstruction model
• Adapts to various undersampling patterns
• Improves image resolution and speed

Innovation

• Unified model for varied undersampling
• Neural operators for resolution flexibility
• Enhanced speed and image quality