EUVIMEDEUVIMED entdecken
Crossref

Dual-branch diffusion model with enhanced mamba for MRI blind super resolution reconstruction

Xiebo Qiu, Xiaoqiang Zhao, Zhenrui Peng, Zhaoyang Song

Physica Scripta · 20.03.2026 · 101 (11) · 115003

DOI: 10.1088/1402-4896/ae4e61 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physica Scripta, online erschienen am 17.03.2026. Verfasst von Xiebo Qiu und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 32 Literaturangaben. Gefördert von: Natural Science Foundation of Gansu Province, Industrial Support Project of Education Department of Gansu Province, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Magnetic resonance imaging (MRI) can help experts accurately diagnose various diseases. Due to the patient’s slight breathing and motion during the prolonged scanning time, the acquired MRI images may suffer from low resolution and blurring, which could affect the clinical diagnostic precision. To address this problem, a novel dual-branch diffusion model with enhanced mamba for MRI blind super-resolution reconstruction (DDEM) is proposed. Specifically, to accurately capture fine structures in MRI images, a feature-aware extractor which comprises an LR encoder and a high-frequency enhancement module (HFEM) is designed, where HFEM selectively supplements essential high-frequency information to enable efficient feature fusion. Moreover, a blur kernel degradation predictor (BKD predictor) is designed which leverages a diffusion model to learn unknown blur kernels from low-resolution MRI images. Subsequently, a region-of-interest-guided reconstructor (ROIG reconstructor) is designed that utilizes the LR-encoded features and predicted blur kernels as priors to guide the diffusion model in performing MRI super-resolution reconstruction. To further improve subtle feature fidelity, a perception-aware enhancement mamba (PAE-Mamba) is designed in the ROIG reconstructor. Extensive experimental results demonstrate that our proposed DDEM outperforms several mainstream methods in super-resolution tasks, and achieves superior quantitative metrics and visual quality.

Förderung

  • Natural Science Foundation of Gansu Province
  • Industrial Support Project of Education Department of Gansu Province
  • National Natural Science Foundation of China

Zitieren

APA

Qiu, X., Zhao, X., Peng, Z., & Song, Z. (2026). Dual-branch diffusion model with enhanced mamba for MRI blind super resolution reconstruction. Physica Scripta, 101(11), 115003. https://doi.org/10.1088/1402-4896/ae4e61

Vancouver

Qiu X, Zhao X, Peng Z, Song Z. Dual-branch diffusion model with enhanced mamba for MRI blind super resolution reconstruction. Physica Scripta. 2026;101(11):115003. doi:10.1088/1402-4896/ae4e61

BibTeX
@article{Qiu2026,
  author = {Qiu, Xiebo and Zhao, Xiaoqiang and Peng, Zhenrui and Song, Zhaoyang},
  title = {Dual-branch diffusion model with enhanced mamba for MRI blind super resolution reconstruction},
  journal = {Physica Scripta},
  year = {2026},
  volume = {101},
  number = {11},
  pages = {115003},
  doi = {10.1088/1402-4896/ae4e61}
}
Kostenlos starten →