Abstract:Magnetic Resonance Imaging (MRI) is one of the most important medical imaging modalities as it provides superior resolution of soft tissues, albeit with a notable limitation in scanning speed. The advent of deep learning has catalyzed the development of cutting-edge methods for the expedited reconstruction of MRI scans, utilizing convolutional neural networks and, more recently, vision transformers. Recently proposed structured state space models (e.g., Mamba) have gained some traction due to their efficiency and low computational requirements compared to transformer models. We propose an innovative MRI reconstruction framework that employs structured state space models at its core, aimed at amplifying both long-range contextual sensitivity and reconstruction efficacy. Comprehensive experiments on public brain MRI datasets show that our model sets new benchmarks beating state-of-the-art reconstruction baselines. Code will be available (<a class="link-external link-https" href="https://github.com/yilmazkorkmaz1/MambaRecon" rel="external noopener nofollow">this https URL</a>).

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the slow reconstruction speed and high consumption of computational resources in magnetic resonance imaging (MRI). Specifically: 1. **Limitations of scanning speed**: Although MRI has excellent resolution in soft - tissue imaging, its scanning speed is slow and the cost is high. To improve its practicality, especially in clinical settings, MRI technology needs to be accelerated. 2. **Reconstruction problems of undersampled k - space data**: Reducing the sampling points in k - space (i.e., Fourier coefficients) can accelerate the scanning speed to a certain extent, but this will make the reconstruction problem ill - posed, that is, there are infinitely many possible solutions. Therefore, effective regularization methods are required to ensure the quality of the reconstructed image. 3. **Limitations of existing methods**: - **Compressed Sensing (CS) method**: Although effective, it is very sensitive to hyper - parameters, and its iterative nature makes it computationally inefficient. - **Convolutional neural networks (CNNs)**: Although they perform well in image processing tasks, their receptive fields are limited and it is difficult to capture long - distance dependencies. - **Transformer - based methods**: Although they can enhance long - distance dependencies, the time complexity of the self - attention mechanism is quadratic, resulting in an excessive computational burden. 4. **Introduction of Structured State Space Models (SSMs)**: To solve the above problems, the author proposes a new MRI reconstruction framework, using SSMs as the core component, aiming to improve long - distance context sensitivity and reconstruction effect while maintaining low computational complexity. ### Specific objectives - **Improve reconstruction quality**: By introducing SSMs, enhance the model's sensitivity to long - distance dependencies, thereby improving the quality of the reconstructed image. - **Reduce computational complexity**: Compared with existing Transformer models, SSMs have a linear time complexity and can process long - sequence data more efficiently. - **Combine physical models**: By alternately using data consistency blocks and SSM blocks, ensure that the reconstruction process conforms to the physical model, improving robustness and accuracy. In summary, this paper aims to develop a lightweight and efficient MRI reconstruction framework by introducing structured state space models to address the shortcomings of existing methods in terms of reconstruction quality and computational efficiency.

MambaRecon: MRI Reconstruction with Structured State Space Models

Deep Learning Based MRI Reconstruction with Transformer

PILOT: Physics-Informed Learned Optimized Trajectories for Accelerated MRI

MRI Reconstruction with Interpretable Pixel-Wise Operations Using Reinforcement Learning

ReconFormer: Accelerated MRI Reconstruction Using Recurrent Transformer

A Comprehensive Survey of Mamba Architectures for Medical Image Analysis: Classification, Segmentation, Restoration and Beyond

Vision Mamba: Cutting-Edge Classification of Alzheimer's Disease with 3D MRI Scans

Machine Learning in Magnetic Resonance Imaging: Image Reconstruction

Transform Learning for Magnetic Resonance Image Reconstruction: From Model-based Learning to Building Neural Networks

MMR-Mamba: Multi-Modal MRI Reconstruction with Mamba and Spatial-Frequency Information Fusion

Deep Convolutional Encoder-Decoder algorithm for MRI brain reconstruction

Multi-branch Cascaded Swin Transformers with Attention to k-space Sampling Pattern for Accelerated MRI Reconstruction

A Systematic Review and Identification of the Challenges of Deep Learning Techniques for Undersampled Magnetic Resonance Image Reconstruction

ERNAS: An Evolutionary Neural Architecture Search for Magnetic Resonance Image Reconstructions

MA-RECON: Mask-aware deep-neural-network for robust fast MRI k-space interpolation

Deep MRI Reconstruction: Unrolled Optimization Algorithms Meet Neural Networks

AVS-Net: Attention-based Variable Splitting Network for P-MRI Acceleration.

SegMamba: Long-range Sequential Modeling Mamba For 3D Medical Image Segmentation

MD-Recon-Net: A Parallel Dual-Domain Convolutional Neural Network for Compressed Sensing MRI

HUMUS-Net: Hybrid unrolled multi-scale network architecture for accelerated MRI reconstruction

Soft Masked Mamba Diffusion Model for CT to MRI Conversion