BM3D-based ultrasound image denoising via brushlet thresholding

Yu Gan, Elsa Angelini, Andrew Laine, Christine Hendon

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

22 Scopus citations

Abstract

In this paper, we present a brushlet-based block matching 3D (BM3D) method to collaboratively denoise ultrasound images. Through dividing image into multiple blocks, we group them based on similarity. Then, grouped blocks sharing similarity form a 3D image volume. For each volume, brushlet thresholding is applied to remove noise in the frequency domain. Upon completion of individual filtering, the volumes are aggregated and reconstructed globally. To evaluate our method, we run our denoising scheme on synthetic images corrupted with additive or multiplicative noise. The results show that our method can achieve good denoising performance in comparison with existing methods. Our method is also evaluated on cardiac and fetal ultrasound images. Analysis on the contrast and homogeneity of the denoised images demonstrates the feasibility of applying our method to ultrasound images to improve image quality and facilitate further processing such as segmentation.

Original languageEnglish
Title of host publication2015 IEEE 12th International Symposium on Biomedical Imaging, ISBI 2015
Pages667-670
Number of pages4
ISBN (Electronic)9781479923748
DOIs
StatePublished - 21 Jul 2015
Event12th IEEE International Symposium on Biomedical Imaging, ISBI 2015 - Brooklyn, United States
Duration: 16 Apr 201519 Apr 2015

Publication series

NameProceedings - International Symposium on Biomedical Imaging
Volume2015-July
ISSN (Print)1945-7928
ISSN (Electronic)1945-8452

Conference

Conference12th IEEE International Symposium on Biomedical Imaging, ISBI 2015
Country/TerritoryUnited States
CityBrooklyn
Period16/04/1519/04/15

Keywords

  • BM3D
  • Brushlet
  • image denoising
  • ultrasound imaging

Fingerprint

Dive into the research topics of 'BM3D-based ultrasound image denoising via brushlet thresholding'. Together they form a unique fingerprint.

Cite this