Integrating Local Precision With Global Consistency for Unsupervised Magnetic Resonance Image Registration
Magnetic Resonance in Medicine · 2026-08 · 96 (2) · 908-917
DOI: 10.1002/mrm.70357 ↗Kurz gefasst
Fachartikel in der Zeitschrift Magnetic Resonance in Medicine, online erschienen am 26.03.2026. Verfasst von He Deng und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan, Hubei, China. Hauptthemen laut MEDLINE: Magnetic Resonance Imaging, Brain, Image Processing, Computer-Assisted. Die Arbeit nennt 28 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Xinjiang Production and Construction Corps, Financial Science & Technology Plan Program of the Eighth Division.
Abstract (Originalsprache)
Purpose
To address the issues of interpretability and learning efficiency in traditional single‐stream frameworks, as well as the limitations in capturing long‐range dependencies and semantic relationships caused by the dependence of two‐stream architectures.
Methods
This paper proposes GLACF, a triple‐stream global‐to‐local attention framework for deformable MR image registration, integrating CNNs, Transformers, and coarse‐to‐fine strategies to enhance alignment accuracy. Its key innovations include: a triple‐stream feature extraction based on Transformers for enriched feature representation; multi‐scale decoupling blocks (MSDBs) that leverage low‐resolution features to generate coarse displacement fields; and a voxel‐wise local attention module (VLFM) that refines high‐resolution features to produce precise displacement fields.
Results
Experimental results on three public 3D brain MRI datasets (e.g., LONI LBPA40, IXI, and OASIS) demonstrate that GLACF consistently outperforms state‐of‐the‐art methods in registration similarity, smoothness, and invertibility. Specifically, on LONI LBPA40, GLACF achieves a Dice similarity coefficient (DSC) of 72.7%, structural similarity index measure (SSIM) of 0.973, percentage of negative Jacobian determinants (% of | J ϕ | ≤ 0) of 0.15, and 95th percentile Hausdorff distance (HD95) of 6.088. On IXI, it attains 76.8% DSC, 0.952 SSIM, 0.48% of | J ϕ | ≤ 0, and 3.284 HD95. On OASIS, it reaches 89.1% DSC, 0.973 SSIM, 0.55% of | J ϕ | ≤ 0, and 1.328 HD95.
Conclusion
These results confirm the robustness and accuracy of GLACF across diverse datasets, highlighting its strong potential for clinical translation in high‐fidelity image registration scenarios.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- School of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan, Hubei, China.
- Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System, Wuhan University of Science and Technology, Wuhan, Hubei, China.
- Department of Radiology, First Affiliated Hospital of Shihezi University, Shihezi, Xinjiang, China.
- Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, Huebi, China.
ORCID-Profile: He Deng
Förderung
- National Natural Science Foundation of China
- Xinjiang Production and Construction Corps
- Financial Science & Technology Plan Program of the Eighth Division
- Doctoral Fund Program of the First Affiliated Hospital of Shihezi University
- Tianshan Talent Program for High-level Medical Professionals
- Beijing Medical Award Foundation
Zitieren
APA
Deng, H., Wang, J., Yin, X., & Min, X. (2026). Integrating Local Precision With Global Consistency for Unsupervised Magnetic Resonance Image Registration. Magnetic Resonance in Medicine, 96(2), 908-917. https://doi.org/10.1002/mrm.70357
Vancouver
Deng H, Wang J, Yin X, Min X. Integrating Local Precision With Global Consistency for Unsupervised Magnetic Resonance Image Registration. Magnetic Resonance in Medicine. 2026;96(2):908-917. doi:10.1002/mrm.70357
BibTeX
@article{Deng2026,
author = {Deng, He and Wang, Jiangfeng and Yin, Xi and Min, Xiangde},
title = {Integrating Local Precision With Global Consistency for Unsupervised Magnetic Resonance Image Registration},
journal = {Magnetic Resonance in Medicine},
year = {2026},
volume = {96},
number = {2},
pages = {908--917},
doi = {10.1002/mrm.70357}
}
EUVIMED