Advancing Quantitative Susceptibility Mapping With 2.5D Diffusion Models for Rapid Intracranial Hemorrhage Quantification
Magnetic Resonance in Medicine · 2026-08 · 96 (2) · 596-610
DOI: 10.1002/mrm.70358 ↗Kurz gefasst
Fachartikel in der Zeitschrift Magnetic Resonance in Medicine, online erschienen am 24.03.2026. Verfasst von Zhuang Xiong und 6 weiteren Autor:innen. Beteiligte Einrichtung: Image X Institute, Sydney School of Health Sciences, Faculty of Medicine and Health, University of Sydney, Sydney, Australia. Hauptthemen laut MEDLINE: Intracranial Hemorrhages, Imaging, Three-Dimensional, Diffusion Magnetic Resonance Imaging, Image Processing, Computer-Assisted. Die Arbeit nennt 72 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Australian Research Council, National Health and Medical Research Council, National Natural Science Foundation of China.
Abstract (Originalsprache)
Purpose
To develop a generative diffusion model‐based approach for robust and efficient quantitative susceptibility mapping (QSM) reconstruction in intracranial hemorrhage (ICH), applicable to both standard gradient echo (GRE) and rapid echo planar imaging (EPI) acquisitions.
Methods
QSMDiff, an unsupervised diffusion model for 3D QSM dipole inversion, was proposed. Three volumetric partitioning strategies including 2D slices, 3D patches, and 2.5D slabs were evaluated, and the memory‐efficient 2.5D slab approach was adopted to balance accuracy and efficiency while maintaining anatomical fidelity. A conditional sampling mechanism ensured consistency with measured local fields, and a three‐stage training data‐generation strategy combining public dataset, synthetic QSM, and ICH lesions simulated from in vivo patients was implemented to overcome data scarcity.
Results
QSMDiff achieved the best overall performance in simulation studies, with SSIM of 0.97 ± 0.07, RMSE of 0.04 ± 0.03, and HFEN of 4.49 ± 0.83, demonstrating superior structural fidelity and noise suppression. For in vivo ICH patients scanned with rapid EPI, QSMDiff showed strong agreement with SWI‐QSM references ( R 2 = 0.83), producing susceptibility estimates with minimal bias and variance. Qualitative evaluation confirmed enhanced resolution and effective artifact suppression in conditions of low SNR, limited resolution, and motion.
Conclusion
QSMDiff achieves high‐quality and accurate QSM reconstruction from both standard GRE and rapid EPI scans for ICH assessment. By integrating a 2.5D training strategy with synthetic ICH augmentation, it delivers accurate and reliable susceptibility maps even from lower‐quality acquisitions, offering a practical solution for fast and robust ICH assessment.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Image X Institute, Sydney School of Health Sciences, Faculty of Medicine and Health, University of Sydney, Sydney, Australia.
- School of Computer Science and Engineering, Central South University, Changsha, China.
- School of Electrical Engineering and Computer Science, University of Queensland, Brisbane, Australia.
- Department of Biomedical Engineering, University of Alberta, Edmonton, Canada.
- School of Clinical Medicine, University of New South Wales, Sydney, Australia.
- School of Engineering, University of Newcastle, Newcastle, Australia.
ORCID-Profile: Yang Gao
Förderung
- Australian Research Council
- National Health and Medical Research Council
- National Natural Science Foundation of China
- Natural Science Foundation of Hunan
Zitieren
APA
Xiong, Z., Gao, Y., Liu, F., Emery, D., Butcher, K., Wilman, A. H., & Sun, H. (2026). Advancing Quantitative Susceptibility Mapping With 2.5D Diffusion Models for Rapid Intracranial Hemorrhage Quantification. Magnetic Resonance in Medicine, 96(2), 596-610. https://doi.org/10.1002/mrm.70358
Vancouver
Xiong Z, Gao Y, Liu F, Emery D, Butcher K, Wilman AH, et al. Advancing Quantitative Susceptibility Mapping With 2.5D Diffusion Models for Rapid Intracranial Hemorrhage Quantification. Magnetic Resonance in Medicine. 2026;96(2):596-610. doi:10.1002/mrm.70358
BibTeX
@article{Xiong2026,
author = {Xiong, Zhuang and Gao, Yang and Liu, Feng and Emery, Derek and Butcher, Ken and Wilman, Alan H and Sun, Hongfu},
title = {Advancing Quantitative Susceptibility Mapping With 2.5D Diffusion Models for Rapid Intracranial Hemorrhage Quantification},
journal = {Magnetic Resonance in Medicine},
year = {2026},
volume = {96},
number = {2},
pages = {596--610},
doi = {10.1002/mrm.70358}
}
EUVIMED