Denoising magnetic resonance images using collaborative non-local means

Geng Chen, Pei Zhang, Yafeng Wu, Dinggang Shen, Pew Thian Yap

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

Noise artifacts in magnetic resonance (MR) images increase the complexity of image processing workflows and decrease the reliability of inferences drawn from the images. It is thus often desirable to remove such artifacts beforehand for more robust and effective quantitative analysis. It is important to preserve the integrity of relevant image information while removing noise in MR images. A variety of approaches have been developed for this purpose, and the non-local means (NLM) filter has been shown to be able to achieve state-of-the-art denoising performance. For effective denoising, NLM relies heavily on the existence of repeating structural patterns, which however might not always be present within a single image. This is especially true when one considers the fact that the human brain is complex and contains a lot of unique structures. In this paper we propose to leverage the repeating structures from multiple images to collaboratively denoise an image. The underlying assumption is that it is more likely to find repeating structures from multiple scans than from a single scan. Specifically, to denoise a target image, multiple images, which may be acquired from different subjects, are spatially aligned to the target image, and an NLM-like block matching is performed on these aligned images with the target image as the reference. This will significantly increase the number of matching structures and thus boost the denoising performance. Experiments on both synthetic and real data show that the proposed approach, collaborative non-local means (CNLM), outperforms the classic NLM and yields results with markedly improved structural details.

Original languageEnglish
Pages (from-to)215-227
Number of pages13
JournalNeurocomputing
Volume177
DOIs
Publication statusPublished - 2016 Feb 12

Fingerprint

Magnetic resonance
Artifacts
Noise
Magnetic Resonance Spectroscopy
Workflow
Brain
Image processing
Chemical analysis
Experiments

Keywords

  • Block matching
  • Denoising
  • Non-local means
  • Non-parametric regression

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Cognitive Neuroscience

Cite this

Denoising magnetic resonance images using collaborative non-local means. / Chen, Geng; Zhang, Pei; Wu, Yafeng; Shen, Dinggang; Yap, Pew Thian.

In: Neurocomputing, Vol. 177, 12.02.2016, p. 215-227.

Research output: Contribution to journalArticle

Chen, Geng ; Zhang, Pei ; Wu, Yafeng ; Shen, Dinggang ; Yap, Pew Thian. / Denoising magnetic resonance images using collaborative non-local means. In: Neurocomputing. 2016 ; Vol. 177. pp. 215-227.
@article{32d844aa5dbd464184771e910f3f3c0b,
title = "Denoising magnetic resonance images using collaborative non-local means",
abstract = "Noise artifacts in magnetic resonance (MR) images increase the complexity of image processing workflows and decrease the reliability of inferences drawn from the images. It is thus often desirable to remove such artifacts beforehand for more robust and effective quantitative analysis. It is important to preserve the integrity of relevant image information while removing noise in MR images. A variety of approaches have been developed for this purpose, and the non-local means (NLM) filter has been shown to be able to achieve state-of-the-art denoising performance. For effective denoising, NLM relies heavily on the existence of repeating structural patterns, which however might not always be present within a single image. This is especially true when one considers the fact that the human brain is complex and contains a lot of unique structures. In this paper we propose to leverage the repeating structures from multiple images to collaboratively denoise an image. The underlying assumption is that it is more likely to find repeating structures from multiple scans than from a single scan. Specifically, to denoise a target image, multiple images, which may be acquired from different subjects, are spatially aligned to the target image, and an NLM-like block matching is performed on these aligned images with the target image as the reference. This will significantly increase the number of matching structures and thus boost the denoising performance. Experiments on both synthetic and real data show that the proposed approach, collaborative non-local means (CNLM), outperforms the classic NLM and yields results with markedly improved structural details.",
keywords = "Block matching, Denoising, Non-local means, Non-parametric regression",
author = "Geng Chen and Pei Zhang and Yafeng Wu and Dinggang Shen and Yap, {Pew Thian}",
year = "2016",
month = "2",
day = "12",
doi = "10.1016/j.neucom.2015.11.031",
language = "English",
volume = "177",
pages = "215--227",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",

}

TY - JOUR

T1 - Denoising magnetic resonance images using collaborative non-local means

AU - Chen, Geng

AU - Zhang, Pei

AU - Wu, Yafeng

AU - Shen, Dinggang

AU - Yap, Pew Thian

PY - 2016/2/12

Y1 - 2016/2/12

N2 - Noise artifacts in magnetic resonance (MR) images increase the complexity of image processing workflows and decrease the reliability of inferences drawn from the images. It is thus often desirable to remove such artifacts beforehand for more robust and effective quantitative analysis. It is important to preserve the integrity of relevant image information while removing noise in MR images. A variety of approaches have been developed for this purpose, and the non-local means (NLM) filter has been shown to be able to achieve state-of-the-art denoising performance. For effective denoising, NLM relies heavily on the existence of repeating structural patterns, which however might not always be present within a single image. This is especially true when one considers the fact that the human brain is complex and contains a lot of unique structures. In this paper we propose to leverage the repeating structures from multiple images to collaboratively denoise an image. The underlying assumption is that it is more likely to find repeating structures from multiple scans than from a single scan. Specifically, to denoise a target image, multiple images, which may be acquired from different subjects, are spatially aligned to the target image, and an NLM-like block matching is performed on these aligned images with the target image as the reference. This will significantly increase the number of matching structures and thus boost the denoising performance. Experiments on both synthetic and real data show that the proposed approach, collaborative non-local means (CNLM), outperforms the classic NLM and yields results with markedly improved structural details.

AB - Noise artifacts in magnetic resonance (MR) images increase the complexity of image processing workflows and decrease the reliability of inferences drawn from the images. It is thus often desirable to remove such artifacts beforehand for more robust and effective quantitative analysis. It is important to preserve the integrity of relevant image information while removing noise in MR images. A variety of approaches have been developed for this purpose, and the non-local means (NLM) filter has been shown to be able to achieve state-of-the-art denoising performance. For effective denoising, NLM relies heavily on the existence of repeating structural patterns, which however might not always be present within a single image. This is especially true when one considers the fact that the human brain is complex and contains a lot of unique structures. In this paper we propose to leverage the repeating structures from multiple images to collaboratively denoise an image. The underlying assumption is that it is more likely to find repeating structures from multiple scans than from a single scan. Specifically, to denoise a target image, multiple images, which may be acquired from different subjects, are spatially aligned to the target image, and an NLM-like block matching is performed on these aligned images with the target image as the reference. This will significantly increase the number of matching structures and thus boost the denoising performance. Experiments on both synthetic and real data show that the proposed approach, collaborative non-local means (CNLM), outperforms the classic NLM and yields results with markedly improved structural details.

KW - Block matching

KW - Denoising

KW - Non-local means

KW - Non-parametric regression

UR - http://www.scopus.com/inward/record.url?scp=84959526680&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84959526680&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2015.11.031

DO - 10.1016/j.neucom.2015.11.031

M3 - Article

VL - 177

SP - 215

EP - 227

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -