CHARMS: A CNN-Transformer Hybrid with Attention Regularization for MRI Super-Resolution
Sensors · 2026
DOI: 10.3390/s26020738 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, 2026. Verfasst von Xia Li und 2 weiteren Autor:innen. Beteiligte Einrichtung: College of Information Engineering, China Jiliang University, Hangzhou 314423, China. Hauptthemen laut MEDLINE: Magnetic Resonance Imaging, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 99 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Fujian Medical University Start-up Fund for Scientific Research, Joint China–Sweden Mobility program from STINT, Zhejiang Natural Science Foundation of China.
Abstract (Originalsprache)
Magnetic resonance imaging (MRI) super-resolution (SR) enables high-resolution reconstruction from low-resolution acquisitions, reducing scan time and easing hardware demands. However, most deep learning-based SR models are large and computationally heavy, limiting deployment in clinical workstations, real-time pipelines, and resource-restricted platforms such as low-field and portable MRI. We introduce CHARMS, a lightweight convolutional–Transformer hybrid with attention regularization optimized for MRI SR. CHARMS employs a Reverse Residual Attention Fusion backbone for hierarchical local feature extraction, Pixel–Channel and Enhanced Spatial Attention for fine-grained feature calibration, and a Multi-Depthwise Dilated Transformer Attention block for efficient long-range dependency modeling. Novel attention regularization suppresses redundant activations, stabilizes training, and enhances generalization across contrasts and field strengths. Across IXI, Human Connectome Project Young Adult, and paired 3T/7T datasets, CHARMS (~1.9M parameters; ~30 GFLOPs for 256 × 256) surpasses leading lightweight and hybrid baselines (EDSR, PAN, W2AMSN-S, and FMEN) by 0.1–0.6 dB PSNR and up to 1% SSIM at ×2/×4 upscaling, while reducing inference time ~40%. Cross-field fine-tuning yields 7T-like reconstructions from 3T inputs with ~6 dB PSNR and 0.12 SSIM gains over native 3T. With near-real-time performance (~11 ms/slice, ~1.6–1.9 s per 3D volume on RTX 4090), CHARMS offers a compelling fidelity–efficiency balance for clinical workflows, accelerated protocols, and portable MRI.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- College of Information Engineering, China Jiliang University, Hangzhou 314423, China.
- School of Medical Imaging, Fujian Medical University, Fuzhou 350005, China.
- Department of Medical Radiation Physics and Nuclear Medicine, Karolinska University Hospital, 171 76 Stockholm, Sweden.
- Department of Clinical Science, Intervention and Technology, Karolinska Institutet, 171 77 Stockholm, Sweden.
ORCID-Profile: Xia Li, Tie-Qiang Li
Förderung
- Fujian Medical University Start-up Fund for Scientific Research
- Joint China–Sweden Mobility program from STINT
- Zhejiang Natural Science Foundation of China
Zitieren
APA
Li, X., Sun, H., & Li, T. Q. (2026). CHARMS: A CNN-Transformer Hybrid with Attention Regularization for MRI Super-Resolution. Sensors. https://doi.org/10.3390/s26020738
Vancouver
Li X, Sun H, Li TQ. CHARMS: A CNN-Transformer Hybrid with Attention Regularization for MRI Super-Resolution. Sensors. 2026. doi:10.3390/s26020738
BibTeX
@article{Li2026,
author = {Li, Xia and Sun, Haicheng and Li, Tie-Qiang},
title = {CHARMS: A CNN-Transformer Hybrid with Attention Regularization for MRI Super-Resolution},
journal = {Sensors},
year = {2026},
doi = {10.3390/s26020738}
}
EUVIMED