Skip to main navigation Skip to search Skip to main content

A new multiscale Bayesian algorithm for speckle reduction in medical ultrasound images

  • K.N. Toosi University of Technology
  • Université de Picardie Jules Verne
  • Tehran University of Medical Sciences

Research output: Contribution to journal › Journal Article › peer-review

16 Citations (Scopus)

Abstract

This paper introduces a new multiscale speckle reduction method based on the extraction of wavelet interscale dependencies to visually enhance the medical ultrasound images and improve clinical diagnosis. The logarithm of the image is first transformed to the oriented dual-tree complex wavelet domain. It is then shown that the adjacent subband coefficients of the log-transformed ultrasound image can be successfully modeled using the general form of bivariate isotropic stable distributions, while the speckle coefficients can be approximated using a zero-mean bivariate Gaussian model. Using these statistical models, we design a new discrete bivariate Bayesian estimator based on minimizing the mean square error (MSE). To assess the performance of the proposed method, four image quality metrics, namely signal-to-noise ratio, MSE, coefficient of correlation, and edge preservation index, were computed on 80 medical ultrasound images. Moreover, a visual evaluation was carried out by two medical experts. The numerical results indicated that the new method outperforms the standard spatial despeckling filters, homomorphic Wiener filter, and new multiscale speckle reduction methods based on generalized Gaussian and symmetric alpha-stable priors.

Original languageEnglish
Pages (from-to)359-375
Number of pages17
JournalSignal, Image and Video Processing
Volume4
Issue number3
DOIs
Publication statusPublished - 2010
Externally publishedYes

Keywords

  • Bivariate minimum mean squared error (MMSE) estimator
  • Bivariate stable distributions
  • Oriented dual-tree complex wavelet transform (DTCWT)
  • Speckle noise

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'A new multiscale Bayesian algorithm for speckle reduction in medical ultrasound images'. These topics are generated from the title and abstract of the publication. Together, they form a unique fingerprint.

Cite this