HiFi-Mamba: dual-stream w-Laplacian enhanced Mamba for high-fidelity MRI reconstruction
HiFi-Mamba: dual-stream w-Laplacian enhanced Mamba for high-fidelity MRI reconstruction
Reconstructing high-fidelity MR images from undersampled k-space data remains a challenging problem in MRI. While Mamba variants for vision tasks offer promising long-range modeling capabilities with linear-time complexity, their direct application to MRI reconstruction inherits two key limitations: (1) insensitivity to high-frequency anatomical details; and (2) reliance on redundant multi-directional scanning. To address these limitations, we introduce High-Fidelity Mamba (HiFi-Mamba), a novel dual-stream Mamba-based architecture comprising stacked w-Laplacian (WL) and HiFi-Mamba blocks. Specifically, the WL block performs fidelity-preserving spectral decoupling, producing complementary low-and high-frequency streams. This separation enables the HiFi-Mamba block to focus on low-frequency structures, enhancing global feature modeling. Concurrently, the HiFi-Mamba block selectively integrates high-frequency features through adaptive state-space modulation, preserving comprehensive spectral details. To eliminate the scanning redundancy, the HiFi-Mamba block adopts a streamlined unidirectional traversal strategy that preserves long-range modeling capability with improved computational efficiency. Extensive experiments on standard MRI reconstruction benchmarks demonstrate that HiFi-Mamba consistently outperforms state-of-the-art CNN-based, Transformer-based, and other Mamba-based models in reconstruction accuracy while maintaining a compact and efficient model design.
2868-2876
Chen, Hongli
3980d1d6-2d2d-442d-81a8-54488aee6aaa
Fang, Pengcheng
7f3b5cc1-6fd3-4e94-8338-0820f3fbd189
Chen, Yuxia
95cf44fa-5048-4bdc-9a68-a275338e632d
Ren, Yingxuan
ec4fc23e-013b-4e00-b711-1bf36355542d
Hao, Jing
edc4198a-0fda-4538-ba41-cf71fd21d3c2
Tang, Fangfang
774813e9-d906-43b1-9d30-16ad10a2259a
Cai, Xiaohao
de483445-45e9-4b21-a4e8-b0427fc72cee
Shan, Shanshan
fb7e00e7-30c5-42e9-8430-dfa31337b925
Liu, Feng
30a0aea2-0008-4d2b-bfb9-28527884967c
14 March 2026
Chen, Hongli
3980d1d6-2d2d-442d-81a8-54488aee6aaa
Fang, Pengcheng
7f3b5cc1-6fd3-4e94-8338-0820f3fbd189
Chen, Yuxia
95cf44fa-5048-4bdc-9a68-a275338e632d
Ren, Yingxuan
ec4fc23e-013b-4e00-b711-1bf36355542d
Hao, Jing
edc4198a-0fda-4538-ba41-cf71fd21d3c2
Tang, Fangfang
774813e9-d906-43b1-9d30-16ad10a2259a
Cai, Xiaohao
de483445-45e9-4b21-a4e8-b0427fc72cee
Shan, Shanshan
fb7e00e7-30c5-42e9-8430-dfa31337b925
Liu, Feng
30a0aea2-0008-4d2b-bfb9-28527884967c
Chen, Hongli, Fang, Pengcheng, Chen, Yuxia, Ren, Yingxuan, Hao, Jing, Tang, Fangfang, Cai, Xiaohao, Shan, Shanshan and Liu, Feng
(2026)
HiFi-Mamba: dual-stream w-Laplacian enhanced Mamba for high-fidelity MRI reconstruction.
Koenig, Sven, Jenkins, Chad and Taylor, Matthew E.
(eds.)
In AAAI-26 Technical Tracks 4.
vol. 40,
AAAI Press.
.
(doi:10.1609/aaai.v40i4.37277).
Record type:
Conference or Workshop Item
(Paper)
Abstract
Reconstructing high-fidelity MR images from undersampled k-space data remains a challenging problem in MRI. While Mamba variants for vision tasks offer promising long-range modeling capabilities with linear-time complexity, their direct application to MRI reconstruction inherits two key limitations: (1) insensitivity to high-frequency anatomical details; and (2) reliance on redundant multi-directional scanning. To address these limitations, we introduce High-Fidelity Mamba (HiFi-Mamba), a novel dual-stream Mamba-based architecture comprising stacked w-Laplacian (WL) and HiFi-Mamba blocks. Specifically, the WL block performs fidelity-preserving spectral decoupling, producing complementary low-and high-frequency streams. This separation enables the HiFi-Mamba block to focus on low-frequency structures, enhancing global feature modeling. Concurrently, the HiFi-Mamba block selectively integrates high-frequency features through adaptive state-space modulation, preserving comprehensive spectral details. To eliminate the scanning redundancy, the HiFi-Mamba block adopts a streamlined unidirectional traversal strategy that preserves long-range modeling capability with improved computational efficiency. Extensive experiments on standard MRI reconstruction benchmarks demonstrate that HiFi-Mamba consistently outperforms state-of-the-art CNN-based, Transformer-based, and other Mamba-based models in reconstruction accuracy while maintaining a compact and efficient model design.
This record has no associated files available for download.
More information
Published date: 14 March 2026
Additional Information:
Publisher Copyright:
© 2026, Association for the Advancement of Artificial Intelligence. All rights reserved.
Venue - Dates:
40th AAAI Conference on Artificial Intelligence, AAAI 2026, , Singapore, Singapore, 2026-01-20 - 2026-01-27
Identifiers
Local EPrints ID: 513816
URI: https://eprints.soton.ac.uk/id/eprint/513816
ISSN: 2159-5399
PURE UUID: 8f9d5e00-1f8f-4271-abc4-7244f2796ce6
Catalogue record
Date deposited: 26 Aug 2026 16:50
Last modified: 13 Sep 2026 15:50
Export record
Altmetrics
Contributors
Author:
Hongli Chen
Author:
Pengcheng Fang
Author:
Yuxia Chen
Author:
Yingxuan Ren
Author:
Jing Hao
Author:
Fangfang Tang
Author:
Xiaohao Cai
Author:
Shanshan Shan
Author:
Feng Liu
Editor:
Sven Koenig
Editor:
Chad Jenkins
Editor:
Matthew E. Taylor
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics