EUVIMEDEUVIMED entdecken
Crossref

Three unbiased anisotropic diffusion filtering models with optimized parameters for Rician noise removing in MR images

Samir Yesli, Moussa Diaf

Magnetic Resonance Imaging · 2026-07 · 130 · 110673

DOI: 10.1016/j.mri.2026.110673 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Magnetic Resonance Imaging, 2026-07. Verfasst von Samir Yesli und 1 weiteren Autor:innen. Beteiligte Einrichtung: Laboratoire Vision Artificielle et Automatique des Systèmes (LVAAS), Mouloud Mammeri University (UMMTO), Tizi-Ouzou, B.P.17 RP, 15000, Algeria. Elect…. Hauptthemen laut MEDLINE: Image Enhancement, Magnetic Resonance Imaging, Image Processing, Computer-Assisted, Artifacts. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 67 Literaturangaben.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Due to the inherent correlation between Rice's noise and the Magnetic Resonance (MR) signal that occurs during the MR images formation, denoising these images is an important but complex process, as it can prove insufficient and even result in the loss of information details. To overcome these two drawbacks, filtering methods based on Partial Differential Equations (PDEs) have been introduced, but the parameters of the filters they propose must be determined with great accuracy. The work proposed in this paper deals with this issue. In fact, in order to denoise MR images with high Rice's noise levels, we propose adapted filtering anisotropic diffusion PDE filter models by using precise local parameters adjustment. These filtering models incorporate a robust Rician noise estimator and an unbiased variant of filters based on anisotropic diffusion, whose parameters are all calculated simultaneously and optimized adaptively through the introduction of the PSO algorithm and the use of reference-based and no-reference quality metrics (PSNR, SSIM and Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE)). The various tests performed on synthetic and real MR images provided excellent results compared to some other published methods, thus demonstrating the high performance of the proposed filters in terms of noise reduction and preservation of important details.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Image EnhancementMagnetic Resonance ImagingImage Processing, Computer-AssistedArtifactsDiffusion Magnetic Resonance ImagingImage Interpretation, Computer-AssistedAlgorithmsAnisotropySignal-To-Noise RatioHumansReproducibility of ResultsComputer Simulation

Stichwörter der Autor:innen

Anisotropic diffusionImage Quality Assessment (IQA)Image enhancementMagnetic Resonance Imaging (MRI) image denoisingPartial Differential Equation (PDE)Particle Swarm Optimization (PSO) algorithmRician noise estimator

Autor:innen und Institutionen

  • Laboratoire Vision Artificielle et Automatique des Systèmes (LVAAS), Mouloud Mammeri University (UMMTO), Tizi-Ouzou, B.P.17 RP, 15000, Algeria. Electronic address: samir.yesli@ummto.dz.
  • Laboratoire Vision Artificielle et Automatique des Systèmes (LVAAS), Mouloud Mammeri University (UMMTO), Tizi-Ouzou, B.P.17 RP, 15000, Algeria. Electronic address: moussa.diaf@ummto.dz.

Zitieren

APA

Yesli, S., & Diaf, M. (2026). Three unbiased anisotropic diffusion filtering models with optimized parameters for Rician noise removing in MR images. Magnetic Resonance Imaging, 130, 110673. https://doi.org/10.1016/j.mri.2026.110673

Vancouver

Yesli S, Diaf M. Three unbiased anisotropic diffusion filtering models with optimized parameters for Rician noise removing in MR images. Magnetic Resonance Imaging. 2026;130:110673. doi:10.1016/j.mri.2026.110673

BibTeX
@article{Yesli2026,
  author = {Yesli, Samir and Diaf, Moussa},
  title = {Three unbiased anisotropic diffusion filtering models with optimized parameters for Rician noise removing in MR images},
  journal = {Magnetic Resonance Imaging},
  year = {2026},
  volume = {130},
  pages = {110673},
  doi = {10.1016/j.mri.2026.110673}
}
Kostenlos starten →