Robust Simultaneous Multislice MRI Reconstruction Using Deep Generative Priors

Shoujin Huang,Guanxiong Luo,Yuwan Wang,Kexin Yang,Lingyan Zhang,Jingzhe Liu,Hua Guo,Min Wang,Mengye Lyu
2024-07-31
Abstract:Simultaneous multislice (SMS) imaging is a powerful technique for accelerating magnetic resonance imaging (MRI) acquisitions. However, SMS reconstruction remains challenging due to the complex signal interactions between and within the excited slices. This study presents a robust SMS MRI reconstruction method using deep generative priors. Starting from Gaussian noise, we leverage denoising diffusion probabilistic models (DDPM) to gradually recover the individual slices through reverse diffusion iterations while imposing data consistency from the measured k-space under readout concatenation framework. The posterior sampling procedure is designed such that the DDPM training can be performed on single-slice images without special adjustments for SMS tasks. Additionally, our method integrates a low-frequency enhancement (LFE) module to address a practical issue that SMS-accelerated fast spin echo (FSE) and echo-planar imaging (EPI) sequences cannot easily embed autocalibration signals. Extensive experiments demonstrate that our approach consistently outperforms existing methods and generalizes well to unseen datasets. The code is available at <a class="link-external link-https" href="https://github.com/Solor-pikachu/ROGER" rel="external noopener nofollow">this https URL</a> after the review process.
Image and Video Processing,Artificial Intelligence,Computer Vision and Pattern Recognition,Signal Processing,Medical Physics
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to reconstruct images efficiently and accurately in simultaneous multi - slice (SMS) magnetic resonance imaging (MRI). The SMS technique can significantly reduce the scanning time or increase the slice coverage by using multi - bandwidth (MB) radio - frequency pulses to simultaneously acquire multiple slices. However, the main challenges faced by SMS reconstruction include: 1. **Complex signal interactions**: Due to the simultaneous excitation of multiple slices, complex signal interactions between and within slices are caused, which may lead to artifacts and other image quality problems. 2. **Limitations of traditional methods**: Traditional SMS reconstruction methods, such as parallel - imaging - based techniques, although they solve the problem to a certain extent, still have problems of noise amplification and residual aliasing artifacts. Especially in high - acceleration settings, these problems are more severe. 3. **Lack of self - calibration signals**: For SMS - accelerated fast spin - echo (FSE) and echo - planar imaging (EPI) sequences, it is difficult to embed fully - sampled central k - space signals (self - calibration signals), which makes the reconstruction process more difficult. To address these challenges, the paper proposes a robust SMS MRI reconstruction method based on the denoising diffusion probability model (DDPM) and introduces a low - frequency enhancement module (LFE) to solve the problem of the lack of self - calibration signals in FSE and EPI sequences. This method gradually restores each slice through inverse - diffusion iterations while maintaining data consistency using the measured k - space data. Experimental results show that this method outperforms existing methods on multiple datasets and has good generalization ability.