International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
RESEARCH ARTICLE
OPEN ACCESS
Speckle Reduction Techniques for Ultrasound Images
Palwinder Singh1
1
(Assistant Professor, Department of Computer Science, GNDU, Amritsar, India)
----------------------------------------************************----------------------------------
Abstract:
Images of different body organs play very important role in medical diagnosis. Images can be taken
by using different techniques like x-rays, gamma rays, ultrasound etc. Ultrasound images are widely used
as a diagnosis tool because of its non invasive nature and low cost. The medical images which uses the
principle of coherence suffers from speckle noise, which is multiplicative in nature. Ultrasound images are
coherent images so speckle noise is inherited in ultrasound images which occur at the time of image
acquisition. There are many factors which can degrade the quality of image but noise present in ultrasound
image is a prime factor which can negatively affect result while autonomous machine perception. In this
paper we will discuss types of noises and speckle reduction techniques. In the end, study about speckle
reduction in ultrasound of various researchers will be compared.
Keywords Noise, Speckle, Gaussian, Spatial Filtering, Transform Filtering
----------------------------------------************************---------------------------------the beam forming process and also during the signal
processing stage. Even during the Scan conversion,
I. INTRODUCTION
there could be loss of information due to the
Ultrasound imaging is a medical diagnosis interpolation. So we can say image gets corrupted
technique that uses sound waves of very high with noise during acquisition, transmission, storage
frequency and their echoes. In addition, ultrasound and the retrieval processes. Ultrasound imaging
images have the advantage of being portable, system overview is given below in fig.1.
versatile, and not requiring ionizing radiations [1].
The image generated using ultrasound waves is
called Ultrasonogram. There are many modes of
ultrasound imaging but b-mode and m-mode are
most commonly used methods. Moreover the
diagnosis procedure in ultrasound is of low cost and
in order to diagnose an illness, person need not to
go through dangerous invasive procedures.
Ultrasound images are coherent images so speckle
noise is inherited in ultrasound images which occur
at the time of image acquisition. There are many Fig.1 Ultrasound Imaging System
factors which can degrade the quality of image but
noise present in ultrasound image is a prime factor
which can negatively affect result while II. SPECKLE NOISE
autonomous machine perception [2]. Noise in a
digital image is a very common problem. Noise can Image noise is the random variation of brightness or
be introduced at all stages of Image acquisition. color information in images produced by the sensor
There could be noises due to the loss of proper and circuitry of a scanner or digital camera [3].
contact or air gap between the Transducer probe Speckle is a particular kind of noise which occurs in
and body; there could be noise introduced during
ISSN :2394-2231
http://www.ijctjournal.org
Page 226
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
images obtained by coherent imaging systems like
ultrasound. The coherent imaging in simple terms is
lensless imaging. Speckle noise is a multiplicative
noise which occurs in the coherent imaging, while
other noises are additive noise. Speckle is caused by
interference between coherent waves that,
backscattered by natural surfaces, arrive out of
phase at the sensor. Speckle can be described as
random multiplicative noise. This type of noise is
an inherent property of medical ultrasound imaging.
So, speckle noise reduction is an essential
preprocessing step, whenever ultrasound imaging is
used for medical imaging. The probability
distribution function for speckle noise is given by
gamma distribution,
P( z ) =
z 1
( 1)! a
Peak signal to noise ratio
Let the original noise-free image F(m,n) , noisy
image G(m,n), and the filtered image F(m,n) be
represented where m and n represent the discrete
spatial coordinates of the digital images. Let the
image size be M x N i.e m= 1,2,3M and n=
1,2,3..N
A. MEAN SQUARE ERROR:
For a given image F(m,n), the mean square error of
the image is given as
M N ~
2
MSE = ( F ( m, n ) F ( m, n))
m=1 n =1
B. ROOT MEAN SQUARE:
For a given image F(m,n), the mean square error of
the image is given as
z
a
Where z represents the gray level and variance is
RMSE = MSE
a2. The probability density function of Salt and
C. MEAN ABSOLUTE ERROR:
Pepper noise is graphically represented in figure-2
For a given image F(m,n), the mean square error of
P(z)
the image is given as
M N ~
MAE = F ( m, n) F ( m, n)
m=1 n=1
D. PEAK SIGNAL TO NOISE RATIO:
z
Fig-2 Probability density function of speckle noise
III.
IMAGE QUALITY MEASURES
The quality of an image can be examined
objectively evaluation as well as subjectively. For
subjective evaluation, the image has to be observed
by a human expert [4]. But The human visual
system cannot do pixel by pixel evaluation of given
image, So exact quality of image is difficult to
determine. There are various metrics used for
objective evaluation of an image [4].
Mean square error
Root mean square error
Mean absolute error
ISSN :2394-2231
Peak signal to noise ratio (PSNR) is another
important image metric. It is defined in logarithmic
scale. It is a ratio of peak signal power to noise
power [8]. Since the MSE represents the noise
power and the peak signal power, the PSNR is
defined as:
1
PSNR = 10 * log (
)
10 MSE
There are some other metrics like, universal quality
index (UQI) can be used to evaluate the quality of
an image now-a-days. Further, some parameters,
e.g. method noise and execution time are also used
in literature to evaluate the filtering performance of
a filter.
IV. SPECKLE REDUCTION TECHNIQUES
http://www.ijctjournal.org
Page 227
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
Filter has very important role in image de-noising
process. Using filter technique, in order to decide
particular value of pixel in output image the
neighbor pixels also participate. The values in filter
are known as coefficient rather than pixels. The
filter which we use for denoising is also called as
mask. There are two basic approaches to image denoising, spatial domain filtering methods and
transform domain filtering methods [3,5]. The
spatial filtering process consists simply moving the
filter mask from point to point in an image. At each
point, the response of the filter at that point is
calculated using a predefined relationship. The
filters in frequency domain are more effective than
in spatial domain while reducing noises because it
is to identify noise in frequency domain [6]. When
an image is transformed into the Fourier domain,
the low frequency components usually correspond
to smooth regions or blurred structures of the
image, whereas high-frequency components
represent image details, edges, and noises. Some
standard speckle reduction filters like mean filter,
median filter, lee filter, kuan filter, frost filter are
given below.
A. Mean Filter
It is a traditional method of filtering. A mean filter
[7,8] acts on an image by smoothing it. i.e., it
reduces the variation in terms of intensity between
adjacent pixels. The mean filter is used to suppress
additive noise but edge preservation is not well with
mean filter. The mean filter is a simple moving
window spatial filter, which replaces the center
value in the window with the average of all the
neighbouring pixel values including that centre
value. It is implemented with a convolution mask,
which provides a result that is a weighted sum of
the values of a pixel and its neighbour pixels. It is
also called a linear filter. The mask or kernel is a
square. Often a 3 3 square kernel is used. If the
sum of coefficients of the mask equal to one, then
the average brightness of the image is not changed.
If the sum of the coefficients equal to zero, then
mean filter returns a dark image. Average filter
method is also called neighbourhood average
method. The essential idea of this method is to
replace gray scale value of the center pixel by
average value of neighbourhood pixel gray scale. It
ISSN :2394-2231
is used to reduce AWGN but it can cause blurring
effect. Its filter features are analyzed as follows:
Suppose the noise model for any digital image is
given as
G(x,y) = F (x,y) + N(x,y)
The image after neighborhood smoothing is
1
1
G ( x, y ) =
F ( x, y ) +
N ( x, y )
M ( x, y )S
M ( x, y )S
B. Median Filter
The Median Filter is performed by taking the
magnitude of all of the vectors within a mask and
sorted according to the magnitudes. The pixel with
the median magnitude is then used to replace the
pixel studied [9].The median filter is classified as a
linear filter. It works well to suppress the Salt and
pepper noise. A median filter comes under the class
of nonlinear filter. It also follows the moving
window principle, like mean filter. A 3 3, 5 5, or
7 7 kernel of pixels is moved over the entire image.
First the median of the pixel values in the window
is computed, and then the center pixel of the
window is replaced with the computed median
value. Calculation of Median is done as first sorting
all the pixel values from the surrounding
neighbourhood and then replacing the pixel being
considered with the middle pixel value. It is known
as a rank filter [10]. Median filters exhibit edgepreserving characteristics unlike linear methods
such as average filtering tends to blur edges, which
is very desirable for many image processing
applications as edges contain important information
for segmenting, labelling and preserving detail in
images. It is reasonable to assume that the signal is
of finite length, consisting of samples from F(0) to
F(L-1). If the filters window length is N=2k+1, the
filtering procedure is given by:
G ( n) = Med [ F ( n k )....F ( n ),....., F ( n + k )]
Where G(n) and F(n) are the input and the output
sequences, respectively.
Ultrasound images corrupted with speckle noise can
be processed with mean and median filters. Results
http://www.ijctjournal.org
Page 228
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
of median and mean filter on ultrasound images in the size of the image and Yj is the value of each
pixel in the image.
matlab are given below
=1 M
m 1
(Yi) 2
j=0
If there is no smoothening, the filter will output
only the mean intensity value(Im) of the filter
window. Otherwise, the difference between Cp and
Im is calculated and multiplied with W and then
summed with Im.
The main drawback of Lee filter is that it tends to
ignore speckle noise near edges.
D. Kuan Filter
Kuan filter is a local linear minimum square error
filter based on multiplicative order it does not make
approximation on the noise variance within the
C. Lee Filter
filter window like lee filter it models the
Lee Filter [11] is based on multiplicative speckle multiplicative model of speckle noise into an
model and it can use local statistics to effectively additive linear form [12]. The weighting function
preserve edges. This filter is based on the approach W is computed as follows:
W = (1 C / C ) /(1 + C )
that if the variance over an area is low or constant,
u
i
u
then smoothing will not be performed, otherwise The weighting function is computed from the
smoothing will be performed if variance is estimated noise variation coefficient of the image,
high(near edges).
Cu computed as follows:
Fig.3 Ultrasound denoising using mean and median filter in matlab
Img(i,j)=Im + W*(Cp-Im)
= 1 / ENL
And Ci is the variation coefficient of the image
Where Img is the pixel Value at indices i, j after computed as follows:
C = S / Im
filtering, Im is mean intensity of the filter window,
i
Cp is the center pixel and W is a filter window Where S is the standard deviation in filter window
given by:
and Im is mean intensity value within the window.
2
2
2
The only limitation with Kuan filter is that the ENL
W = /( + )
2
where is the variance of the pixel values within parameter is needed for computation.
the filter window and is calculated as:
2
n 1
= 1 N ( Xj ) 2
j=0
E. Frost Filter
Frost filter is a spatial domain adaptive filter that is
based on multiplicative noise order it adapts to
noise variance within the filter window by applying
Here, N is the size of the filter window and Xj is the exponentially weighting factors M as:
2
pixel value within the filter window at indices j.
M = exp( ( DAMP * ( S / Im) ) * T )
n
The parameter is the additive noise variance of
the image given in following equation, where M is The weighting factor decrease as the variance
within the filter windows reduces. DAMP is a
factor that determines the extent of the exponential
ISSN :2394-2231
http://www.ijctjournal.org
Page 229
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
damping for the image [13]. The larger the damping
value, the heavier is the damping effect. Typically
the value is set to 1. S is the standard deviation of
the filter window, Im is the mean value within the
window and T is the absolute value of the pixel
distance between the center pixel to its surrounding
pixels in the filter window. The value of the filtered
pixel is replaced with a value calculated from
weighted sum of each pixel value Pn and the
weights of each pixel Mn in the filter window over
the total weighted value of the image as:
variance, etc. of a small region or sub-image) are
important [12]. Wiener filtering is based on both the
global statistics and local statistics and is given as
F. Wiener Filter
G. Soft Computing for Ultrasound Despeckling
F ( x, y ) = g +
And
f 2
f 2+ n 2
g =
( g ( x, y ) g )
M
N
g ( s, t )
L s = m t = n
1
Where F ( x, y ) denotes restored image, f2 is the
local variance and n2 is the noise variance [14]. In
statistical theory, Wiener filtering is a great land
Im g (i, j ) = P * M / M
n
n
n
mark. It estimates the original data with minimum
The parameters in the Frost filter are adjusted mean-squared error and hence, the overall noise
according to the local variance in each area. If the power in the filtered output is minimal. Thus, it is
variance is low, then the filtering will cause accepted as a benchmark in 1-D and 2-D signal
extensive smoothing.
processing.
The wiener filter is a spatial domain filter and it Soft computing principles like Artificial Neural
generally used for suppression of additive noise. Networks (ANN), Genetic Algorithms (GA) and
Wiener filters are a class of optimum linear filters Fuzzy Logic (FL) are also be used in designing
which involve linear estimation of a desired signal algorithms for speckle noise reduction in medical
ultrasound images. Hyunkyung Park et al shows
sequence from another related sequence. The
that a cellular neural network which is a kind of
wiener filters main purpose is to reduce the amount
recurrent neural network can deal with images by
of noise present in a image by comparison with an the weight of neurons called a cell. It could obtain
estimation of the desired noiseless image [14]. This more detail image recognition compared with other
filter is the mean squares error-optimal stationary methods. In the study [15], they discuss
linear filter for images degraded by additive noise determination template parameters of the cellular
and blurring. due to linear motion or unfocussed neural network for ultrasound image processing.
optics Wiener filter is the most important technique Their experimental results show effectiveness of
for removal of blur in images. Wiener filter can be applying the proposed method to boundary
applied in two ways (a) spatial domain by using enhancement and the speckle noise reduction of
mean squared method (b) fourier transform method. medical ultrasound image. In [16] Maruyama
Wiener filter in fourier domain can be used for Kenjiro et al presents a neural Network based
deblurring and denoising whereas in spatial domain nonlinear 2D filtering technique for adaptive
Wiener filter cannot be used for deblurring. Wiener speckle reduction in ultrasound images. Then use
filter is based on the least-squared principle, i.e. the ultrasound speckle model and back propogation for
this filter minimizes the mean-squared error (MSE) training the Neural Network. They confirmed the
between the actual output and the desired output. efficiency of the approach with simulated results.
Thus, both global statistics (mean, variance, etc. of
the whole image) and local statistics (mean,
V.
SPECKLE REDUCTION TECHNIQUES AND RELETIVE FINDINGS
ISSN :2394-2231
http://www.ijctjournal.org
Page 230
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
Year
Author
Title
Approach
Relevant Findings
1980
Lee
Digital Image enhancement and
noise filtering
Single scale
It uses the approach that if the variance
over an area is low or constant, then
smoothing will not be performed,
otherwise smoothing will be performed if
variance is high.
1982
Frost et al
A model for radar image & its
application to adaptive digital
filtering for multiplicative noise
Single Scale
It belongs to spatial domain adaptive filter
that works on multiplicative noise , it
adapts to noise variance within the filter
window by using exponentially weighting
factors.
1989
T.Loupas et al
An adaptive weighted median filter
for speckle suppression in medical
ultrasonic images
Single Scale
They do the adaptive filtering by adjusting
the weight coefficients of the median filter
according to the local statistics of the
image.
1995
Richard N.
Czerwinska et al
Ultrasound speckle Reduction by
Directional Median filtering
Single Scale
A technique is presented which uses novel
adaptation of the median filter to the
problem of speckle reduction by
preserving boundary in ultrasonic
imaging.
2001
Chedsada
Chinrungrueng et al
Fast Edge-Preserving Noise
Reduction for Ultrasound Images
Single Scale
It describes a non linear filtering
technique which is based on the least
squares fitting of a polynomial function to
image intensities.
2004
Yu and acton
Generalized speckle reducing
anisotropic diffusion for ultrasound
imagery
Single Scale
PDE for speckle reduction from
minimizing a cost functional of
instantaneous coefficient of variation was
developed.
2006
Badawi et al
Speckle Reduction in Medical
Ultrasound: A Novel Scatterer
Density Weighted Nonlinear
Diffusion Algorithm Implemented
as a Neural-Network Filter
Single scale
They Proposed a novel algorithm for
speckle reduction in medical ultrasound
imaging while preserving edges with
added advantage of adaptive noise
filtering and also speed.
ISSN :2394-2231
http://www.ijctjournal.org
Page 231
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
2007
Ricardo G. Dantas et
al
Ultrasound speckle reduction using
modified gabor filters
Single Scale
It describes a method for speckle
reduction in ultrasound medical imaging,
which uses a bank of wideband 2-D
directive filters, based on modified Gabor
function.
2009
Shankar
Contrast enhancement and phasesensitive boundary detection in
ultrasonic speckle using Bessel
spatial filters
Single Scale
A class of spatial filters based on
cylindrical Bessel functions of the first
kind for speckle reduction was proposed.
2011
Babak
Mohammadzadeh et
al
Contrast Enhancement and
Robustness Improvement of
Adaptive Ultrasound
Imaging Using Forward-Backward
Minimum Variance Beamforming
Single Scale
They used forward/backward spatial
averaging for array covariance matrix
estimation, which is then employed in
minimum variance weight calculation.
2011
Paul liu and dong liu
Filter-based compounded delay
estimation
with application to strain imaging
Single Scale
They developed an approach using a filter
bank to create multiple looks to produce a
compounded motion estimate.
2003
Pizurica et al
A versatile wavelet domain Noise
filtration technique for medical
imaging
Multi scale
They Proposed a robust wavelet domain
method for noise filtering in medical
images. The proposed method adapts itself
to various types of image noise as well as
to the preference of the medical expert.
2004
S. Gupta et al
Wavelet based statistical approach
for speckle reduction in medical
ultrasound images
Multi Scale
The threshold is calculated using simple
standard deviation of noise and the sub
band data of noise free image . K is also
used as a scale parameter.
2007
Zhang et al
Nonlinear diffusion in laplacian
pyramid domain for ultrasonic
speckle reduction
Multi Scale
They presents a Laplacian pyramid-based
nonlinear diffusion (LPND), approach for
reducing speckle noise in medical
Ultrasound imaging .
2010
Maryam
Amirmazlaghani
Two Novel Bayesian Multiscale
Approaches for Speckle
Suppression in SAR
Images
Multi Scale
They developed two new bayesian speckle
suppression approaches in this paper.
They introduced 2D GARCH model and
GARCH- M model
ISSN :2394-2231
http://www.ijctjournal.org
Page 232
International Journal of Computer Techniques - Volume 3 Issue 2, Mar-Apr 2016
2014
S. Kalaivani et al
Condensed anisotropic diffusion
for speckle reduction and
enhancement in ultrasonography
Multi Scale
In this scheme, diffusion matrix is
designed using local coordinate
transformation and the feature broadening
correction term is derived from energy
function.
Table 1 Speckle reduction techniques and relative findings
[6]
VI.
CONCLUSION
[7]
This paper presents a detailed survey of research on
speckle removal methods. We have focused only on
speckle noise which occurs most frequently in
ultrasound images. Speckle Noise with various
noise intensity range from low to high. We have
analysed noise removal algorithms for these noises.
The parameters for this analysis were high level of
noise detection, preserving features and edges, over
smoothness, high contrast image, high density noise,
and mixture of noises. There is lack of uniformity
in how methods are evaluated so it is imprudent to
declare which methods indeed have lowest error
rate with highest noise ratio. Therefore, our analysis
has produced relative performance of methods.
Noise can be removed using single scale filters as
well multi scale filters. Single scale possesses
mathematical simplicity but have the disadvantage
that they introduce blurring effect. To reduce this
blurring effect we can use multi scale filters like
wavelet filter etc because of their multi resolution
property. So, keeping in view, a robust system
should fulfil all the above parameters with multiple
noises removal in a single ultrasound image and in
multiple images.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
REFERENCES
[1]
[2]
[3]
[4]
[5]
Palwinder singh, leena jain, Noise reduction in ultrasound images
using wavelet and spatial filtering techniques, IEEE conference
IMKE-2013, Pg. 57.
Y. Wang and H. Zhou, Total variation wavelet-based medical image
de-noising School of Mathematics, Georgia Institute of Technology,
Atlanta, GA 30332-0430, USA, 2006.
S.Preeti and D.Narmadha,A Survey On Image Denoising
Techniquesinternational journal of computer applications, Volume58No.6, November 2012
B.Nilamani , Development of Some Novel Spatial-Domain and
Transform- Domain Digital Image Filters , Doctoral Dissertation,
National Institute of Technology, Rourkela, India.
Richard, N. Czerwinska, Douglas, L., Jones and William, D. OBrien,
Jr, Ultrasound Speckle reduction by directional median filtering, in
Proc. IEEE ICIP, Vol. 1, pp: 358-361,1995.
Lee, J., Digital image enhancement and Noise filtering using local
statistics. IEEE Trans. Pattern Anal. Mach. Intell., Vol. PAMI-2, No.
2, pp. 165-168, 1980.
Kuan, D. T., Sawchuk, T., Strand, C. and Chavel., Adaptive
restoration of images with speckle, IEEE Trans. Acoust., Speech,
Signal Process., Vol. ASSP-35, No. 3, pp. 373-383,1987.
Frost, V., Stiles, J., Shanmugan, K. and Holtzman, J., A model for
radar images and its application to adaptive digital filtering of
multiplicative Noise", IEEE Trans. Pattern Anal. Mach. Intell., Vol.
PAMI-4, No. 2, pp. 157-166, 1982.
Solbo, S., and Eltoft, T. A Stationary Wavelet-Domain Wiener Filter
for Correlated Speckle, IEEE Transactions on Geoscience and Remote
sensing, Vol. 46, No. 4, pp. 1219-1230, 2008.
Maruyama Kenjiro and Toshi Nishimura, A study for reduction of
speckle noise in Medical Ultrasonic Images using Neural Network
World Congress On Medical Physics And Biomedical Engineering
2006.
L.Gagnon., A. Jouan, Speckle filtering of SAR images:
A
comparative study between complex-wavelet based and standard
filters, In SPIE Proc., 3169, 1997, pp. 8091.
Pizurica, A., Philips, W., Lemahieu, I. and Acheroy, M., A versatile
wavelet domain Noise filtration technique for medical imaging. IEEE
Trans. Med. Imag., Vol.22, No.3, pp. 323-331, 2003.
Zhang. F., Koh. L.M.,.Yoo. Y.M. and Kim. Y., Nonlinear diffusion in
laplacian pyramid domain for ultrasonic speckle reduction, IEEE
Trans. Med. Imag., Vol.26, No.2, pp.200-211, 2007.
Maryam Amirmazlaghani and Hamidreza Amindavar,, Two Novel
Bayesian Multiscale Approaches for Speckle Suppression in SAR
Images, IEEE Trans. on Geoscience and remote sensing, Vol. 48, No.
7, pp. 2980-2993, 2010.
David Sutton, Radiology and Imaging for Medical Students, 7th
Edition. Churchill Livingstone
Bailey and Loves, Short Practice of Surgery, 25th Edition. Edward
Arnold publishers Ltd, UK.
Leena Jain, Palwinder Singh,A historical and qualitative analysis of
different medical imaging techniques, International Journal of
Computer Applications, Vol-107-No 15, December 2014.
Chedsada Chinrungrueng, Aimamorn Suvichakorn, Fast EdgePreserving Noise Reduction for Ultrasound Images, IEEE
Transactions on Nuclear Science, Vol. 48, No. 3, 2001.
R. Coifman and D. Donoho, "Translation invariant de-noising," in
Lecture Notes in Statistics: Wavelets and Statistics, vol. New York:
Springer-Verlag, pp. 125- -150, 1995.
ISSN :2394-2231
http://www.ijctjournal.org
Page 233
