ˉ420ˉ
Medical Ultrasound Images Despeckling Using a Modified Adaptive Weighted
Bilateral Filter Algorithm
DAI Yun, FU Xiaowei*, Chen Li, Tian Jing, Ding Sheng
College of Computer Science and Technology, Wuhan University of Science and Technology. Hubei Province Key
Laboratory of Intelligent Information Processing and Real-time Industrial System, Wuhan, China, 430065
*corresponding author: fxw_wh0409@wust.edu.cn
Abstract In this paper, we concentrate on a hybrid
method of spatial domain and frequency domain. Firstly,
the speckle noise in the low-pass approximation
component is filtered by the new quantum-inspired
bilateral filter, since the low-pass approximation
component of ultrasound images also contains some
speckle noise. The filtered image is treated as
preprocessed image. Then, according to the statistical
properties of medical ultrasound images in wavelet
domain, the noise-free signal and speckle noise are
modeled as generalized Laplace distribution and
Gaussian distribution respectively. The universal wavelet
threshold function is applied to deal with speckle noise in
the filtered image. Differing from conventional bilateral
filter, the proposed method can not only retain
high-contrast features like edges, but also preserve
low-contrast details like textures without introducing
noise. Experiments are applied to both simulated images
and real ultrasound images, where the proposed
algorithm is compared with other six methods. Results
show that the proposed algorithm can perform better in
reducing the speckle noise while well preserving detail
for ultrasonic images.
Key words quantum-inspired, bilateral filter, wavelet
threshold function, ultrasonic images
1 Introduction
In medical images, ultrasonography has been
considered as one of the most popular modalities of
medical imaging system owing to its lack of radiations
and relatively low-cost. However, medical
ultrasonographic images generally suffer from speckle
noise and are of poor visibility.
Early speckle suppression methods involve the use
of adaptive filters in spatial domain, which include Frost
filter, Lee filter, Kuan filter [1], etc.. However, the results
of despeckling indicate that these methods can only
eliminate noise without maintaining image edge
information very well, which makes the image illegible.
The Wiener filter [2] is applied in FFT domain.
Anisotropic diffusion [3] is used for speckle removal.
These are nonlinear filtering techniques for
simultaneously performing contrast enhancement and
noise reduction by using the coefficient of variation [4].
Recently, discrete wavelet transform (DWT) is the
trend for speckle removal. The GenLik method [5] is one
of the most successful wavelet-based technologies, but
the results of despeckling depend on the selection of
parameters, and the capability of speckle suppression is
influenced by the optimum parameters. Tian Jing et al. [6]
have proposed a non-parametric statistical model to
formulate the marginal distribution of wavelet
coefficients. In addition, a maximum a posteriori (MAP)
estimation-based image despeckling method is derived
by incorporating the proposed model into a Bayesian
inference framework. However, this method not only has
a high time complexity, but also has limitations in
despeckling capability. Many of these shortcomings will
be avoided by the use of complex wavelet transform.
Several complex wavelet transforms like the dual-tree
complex wavelet transform (DTCWT) [7] that can
reduce the artifacts of the critically sampled DWT. In the
above mentioned methods, real-valued filters instead of
complex-valued filters are used and due to presence of
redundancy, they are computationally costly as well.
Considering the limitations of the current methods,
Fu et al. [8] proposed a quantum-inspired adaptive
threshold function for speckle reduction in ultrasound
images and an exponential operation is employed to
reconstruct the despeckling images. Compared with
above despeckling methods, the quantum-inspired
Proceedings of the International Workshop
on Modern Science and Technology
October 30-31, 2014 Wuhan, China