Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.00145 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908856891736064 |
|---|---|
| author | Zheng, Kangyuan Cai, Xuan Wang, Jiangqi Fu, Guixing Li, Zhuoshuo Chen, Yazhou Ge, Xinting Qu, Liangqiong Liu, Mengting |
| author_facet | Zheng, Kangyuan Cai, Xuan Wang, Jiangqi Fu, Guixing Li, Zhuoshuo Chen, Yazhou Ge, Xinting Qu, Liangqiong Liu, Mengting |
| contents | Magnetic Resonance Imaging (MRI) is a crucial non-invasive imaging modality. In routine clinical practice, multi-stack thick-slice acquisitions are widely used to reduce scan time and motion sensitivity, particularly in challenging scenarios such as fetal brain imaging. However, the resulting severe through-plane anisotropy compromises volumetric analysis and downstream quantitative assessment, necessitating robust reconstruction of isotropic high-resolution volumes. Implicit neural representation methods, while achieving high quality, suffer from computational inefficiency due to complex network structures. We present M-Gaussian, adapting 3D Gaussian Splatting to MRI reconstruction. Our contributions include: (1) Magnetic Gaussian primitives with physics-consistent volumetric rendering, (2) neural residual field for high-frequency detail refinement, and (3) multi-resolution progressive training. Our method achieves an optimal balance between quality and speed. On the FeTA dataset, M-Gaussian achieves 40.31 dB PSNR while being 14 times faster, representing the first successful adaptation of 3D Gaussian Splatting to multi-stack MRI reconstruction. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_00145 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | M-Gaussian: An Magnetic Gaussian Framework for Efficient Multi-Stack MRI Reconstruction Zheng, Kangyuan Cai, Xuan Wang, Jiangqi Fu, Guixing Li, Zhuoshuo Chen, Yazhou Ge, Xinting Qu, Liangqiong Liu, Mengting Computer Vision and Pattern Recognition Artificial Intelligence Magnetic Resonance Imaging (MRI) is a crucial non-invasive imaging modality. In routine clinical practice, multi-stack thick-slice acquisitions are widely used to reduce scan time and motion sensitivity, particularly in challenging scenarios such as fetal brain imaging. However, the resulting severe through-plane anisotropy compromises volumetric analysis and downstream quantitative assessment, necessitating robust reconstruction of isotropic high-resolution volumes. Implicit neural representation methods, while achieving high quality, suffer from computational inefficiency due to complex network structures. We present M-Gaussian, adapting 3D Gaussian Splatting to MRI reconstruction. Our contributions include: (1) Magnetic Gaussian primitives with physics-consistent volumetric rendering, (2) neural residual field for high-frequency detail refinement, and (3) multi-resolution progressive training. Our method achieves an optimal balance between quality and speed. On the FeTA dataset, M-Gaussian achieves 40.31 dB PSNR while being 14 times faster, representing the first successful adaptation of 3D Gaussian Splatting to multi-stack MRI reconstruction. |
| title | M-Gaussian: An Magnetic Gaussian Framework for Efficient Multi-Stack MRI Reconstruction |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2603.00145 |