International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]

An Efficient Lossless Medical Image compression

using Hybrid Algorithm
C.Priya
Dr.T.Kesavamurthy
M.UmaPriya
Department of Electronics and communication Department of Electronics and communicationDepartment of Electronics and communication
Karpagam college of Engineering,
PSG college of Technology,
Karpagam college of Engineering,
Coimbatore-32.
Coimbatore-04.
Coimbatore-32.
priya_karthikayeni@yahoo.co.in

property in the time domain and in the

frequency
domain..Embedded
Zerotree
Wavelet is very simple and effective
compression algorithm,but many improvements
over Embedded Zerotree Waveletare achieved
by SPIHT algorithm by Amir Said and William
Pearlman .In this SPIHT algorithm ,many wide
sense Zerotrees are effectively found and they
are representing by separation of the tree root
from the tree.From the result of this technique
,this compression algorithm is very efficient
and giving better results.The SPIHT algorithm
creates a pyramid structure based on a wavelet
decomposition of an image.The wavelet
coefficients at the top side of the pyramid has a
strong spatial relationship with their
children.This SPIHT algorithm bases its
efficiency by searching for significant pixels
throughout the entire pyramid structure tree.If
the parent root component having the value
zero means all the child component value
zero.so it easily neglected all the unwanted
components and finally it stores only the correct
information bytes.Lempel Ziv Welch is a
universal
lossless
data
compression
algorithm.In this paper the main focus point is
the usage of hybrid algorithm,which is the
combination of SPIHT algorithm with the
Lempel Ziv Welch compression algorithm for
3D images.This algorithm guarantees that the
original information cannot be changed after
the compression.This Lempel Ziv coding

Abstract- Recently many new algorithms for

image compression based on wavelets have
been developed.This paper gives a detailed
explanation of SPIHT algorithm with the
combination of Lempel Ziv Welch
compression
technique
for
image
compression by MATLAB implementation.
Set
partitioning
in
Hierarchical
trees(SPIHT) is one of the most efficient
algorithm known today. Pyramid structures
have been created by the SPIHT algorithm
based on a wavelet decomposition of an
image. Lempel Ziv Welch is a universal
lossless
data
compression
algorithm
guarantees that the original information can
be exactly reproduced from the compressed
data.The proposed methods have better
compression ratio, computational speed and
good reconstruction quality of the image. To
analysis the proposed lossless methods here,
calculate the performance metrics as
Compression ratio, Mean square error, Peak
signal to Noise ratio.
IndexTerms
LempelZivWelch(LZW),SPIHT,Wavelet

I. INTRODUCTION
Today ,wavelet transform has been widely
applied and developed in image processing and
image compression .DWT is the part of
mathematics that have a best localization

ISBN 978-93-80609-17-1
215

International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]

The 3-level wavelet decomposition of the

spatial orientation tree structure is in
fig.1.Many new algorithm for image
compression based on wavelets have been
recently developed. This is having some key
concepts like (1)ordered bit-plane transmission
of refinement bits.(2)partial ordering of wavelet
coefficients by magnitude with transmission of
ordered by a subset partitioning the SPIHT
algorithm the orders the wavelets coefficients
according to a significant test and stores this
information in three separate lists. They are list
of insignificant sets(LIS), List of significant
pixels(LSP), and the list of insignificant
pixels(LIP).Finally these lists are also
implemented by array structure. This is creating
a four way decomposition that is LL(Low pass
then low pass), LH(Low pass then high pass),
HL(high pass then high pass),and HH(High
pass then high pass) .This process is repeated
until the top of the pyramid is reached. Finally
this is a simple and efficient algorithm with
many unique and desirable properties. Note that
different compression methods were developed
specifically to achieve at least one of those
objectives. What makes SPIHT really
outstanding is that it yields all those qualities
simultaneously.

scheme is universal and this algorithm uses

adaptive approach and fixed size window that is
sliding window principle.
Volumetric medical image standard is
accepted for storing, transmitting and for
viewing.
Discrete
Wavelet
Transform
(DWT)[3-5] is the important tool for analysing
the images.DWT can furnish the multi
resolution image representation. Discrete
Wavelet Transform using SPHIT (Set Partition
in Hierarchical Tree) algorithm provides best
results for compressing the images.Furthur,
lossless arithmetic coding and Lempel Ziv
Welch (LZW) technique is used for
compressing the data stream. Lempel Ziv
Welch is a universal lossless data compression
algorithm In our method, using LZW technique
[8-10] a dictionary has look ahead buffer of
length 255 bytes and size is of 65,535 bytes
has constructed.
Combination of the state of art method
with arithmetic coding and LZW results in
efficient lossless compression. By combining
the two algorithms can achieve to reduce
number of bits during the transmission and
finally get good compression ratio. Moreover
the proposed method is compared with the the
existing method using the compression metrics
as Peak Signal to Noise ratio, Compression
ratio . Here the section II elaborates the
Sequence of compression Techniques; section
A and section B explain the SPIHT algorithm,
Lempel Ziv Welch technique respectively.
Section III explains the proposed work. Section
IV adopts with the results and comparison then
the final section V gives the conclusion of the
proposed work.

Fig 2.Pyramid structure by the four way decomposition

in SPIHT

III.ALGORITHM
A.Analysis of SPIHT Algorithm
B.Lempel-Ziv Welch Algorith

The LIS, LIP and LSP lists were also built

using array structure. These arrays are
especially useful when handling the refinement
pass as these lists can grow or shrink at the run
time. Here a biorthogonal filter 4.4 is used.
Finally by the SPIHT algorithm, is used to

A.Analysis of SPIHT Algorithm

ISBN 978-93-80609-17-1
216

International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]

calculate, the wavelet decomposition, SPIHT

encoding.spiht
decoding
and
wavelet
reconstruction.

separated from the compressed image. The

header has to transmit followed by the streams
of bits. Once the bit stream is received in the
receiver side, it should be decoded by the LZW
decoding algorithm and SPIHT decoding
algorithm. Finally, to reconstruct the image in
the receiver side the inverse bi-orthogonal
wavelet is applied.

B.Analysis of Lempel Ziv Welch Algorithm

This LZW algorithm guarantees that the
original information can be exactly reproduced
from the compressed data. Here combination of
SPIHT algorithm with Lempel Ziv Welch
technique for further compression and finally
saves a lot of bits during transmission. Finally,
getting the good compression ratio for the 3D
dicom medical images.These arrays are
especially useful when handling the refinement
pass as these lists can grow or shrink at the run
time. Here biorthogonal filters 4.4 are used. The
compressed bit stream from the SPHIT
encoding is obtain is entered to the LZW . It is
a dictionary based algorithm or form Code
book for encoding and decoding process. In
essence, the encoder form a dictionary used to
reduce the redundancy based on the previous
occurrence in the dictionary.
III.PROPOSED WORK
In our proposed method in fig 2 shows
the combination of sequence lossless
compression algorithms. Here, Bi-orthogonal
wavelet transform is applied for three level
decomposition of volumetric medical image
datas.TheTransfer syntax user identification
(TSUID) is modified before compression by
SPIHT.After decomposition, the images are
coded using the state of art method SPIHT.The
Bit stream is then encoded by LZW algorithm.
LZW algorithm constructs a dictionary based
on the input bit stream to it. By identifying the
patterns, not the repeated entries, is entered in
the dictionary. For each pattern as an index get
stored in the dictionary. The index is written as
the output compressed bit stream.

Pseudocode:
SPIHT Algorithm steps:
A)Initialization:
#Output m should be calculated from the coefficients;
#Set the LSP as null list ;
#Set the LIP=(a,b) which belongs to H(a,b);
#Set the LIS=(a,b) which belongs to H(a,b);
but D(a,b) not equal to null set.
B)Sorting pass.
B.1 for each entry (a,b) in the LIP do:
B.1.1 Output P (a,b)
B.1.2 If P(a,b)=1 move (a,b) to the LSP
and

Then the DICOM header is then added

to the compressed bit stream. For transmitting,
the header and the image data have to be

ISBN 978-93-80609-17-1
217

International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]

e.In LZW decoding; Code=read the next code

from the encoder
f.output =table [code].

output the sign of C

B.2 for each entry (i,j) in the LIS do:
B.2.1 if the entry is A type then
output P(D(a,b));
if P(D(a,b))=1 then
# for each (m,n) which belongs to O(a,b) do:
* Output P(m,n)
* If P(m,n)=1 then add(m,n) to the LSP and
output sign of C;
* If P(m,n)=0 then add (m,n) to the end of
LIP
# If L(a,b) not equal to null then move (a,b)to
the end of the LIS as entry of type B and go
to step B.2.2
otherwise remove the entry from the LIS;
B.2.2 if the entry is of type B
Output P(L(a,b))
If P(L(a,b))=1then
# add each (m,n) which belongs to O(m,n) to
the end
of the LIS as entry of the type A
# remove (m,n) from the LIS
C)Refinement pass.
For each entry (m,n) in the LSP except those
included
in the last sorting pass , output m th most
significant bit of C
D)Quantization step.
decrement the m by A and go to step B
Notations are as follows:
L(a,b)=D(a,b)-O(a,b)
O(a,b): set of coordinates of the off-spring
(a,b)
D(a,b): set of coordinates of all descendants
(a,b)
H(a,b): set of coordinates of all tree roots in
the
Highest level of the pyramid.
LZW:
a.In the LZW encoding the sequence is given to
the dictionary technique.
b.Initialize the buffer table [0-255] =individual
bytes
c.New entry in the dictionary=get the input
pattern.
d.Encode data = index of the pattern.

IV.RESULTS AND DISCUSSION

To test the proposed hybrid method described
in the previous section, select four grey scale
DICOM medical images. Figure 3 shows that
the three level decomposition of the images
using Bi-orthogonal wavelet transform.

fig 3a, b, c, d shows the samples of three level decomposition

of original image 1,2,3,4 respectively

fig 4a, b, c, d shows the samples of original and

reconstructed image of image 1,2,3,4 respectively.

ISBN 978-93-80609-17-1
218

International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]

Modality(MRI)

If the rate is increased PSNR value should be

increased also elapsed time also increased.LZ
algorithm is the lossless compression technique
so the signal noise ratio cannot be changed for
LZ compression after executing the SPIHT
algorithm.so for both algorithm the PSNR value
is same and the size of the image only modified
that is compressed.

Proposed Method
CR

Slices :Pixel per slice

PSNR
(dB)

Image 1(24:512x512)

8.99

41.01

Image 2(36:512x512)

9.82

43.22

Image 3 (40:256x256)

9.74

42.09

Image 4(50:512x512)

11

44.79

Image 5(12:128x128)

10.11

44

Image 6(30:256x256)
Image 7(39:256x256)

7.04

40.33

9.92

43.99

Image 8(54:256x256)

9.01

42.01

Image 9(46:512x512)

9.8

43

V.CONCLUSION
This paper proposes a lossless DICOM image
compression using wavelet subsequently coded
by SPHIT followed by dictionary based LZW
coding for compressing the bit stream. From
this SPIHT the essential content does not
damaged and for further compression,
combining the SPIHT with LZW algorithm. By
this method good compression ratio and this
method saves a lots of bits during transmission
and giving the good compression ratio. The
proposed hybrid algorithm yields high PSNR
value thus saving time.

Table 1: Comparison of PSNR and CR values for different

medical images

REFERENCES
[1]. D. Salomon, Data Compression: The
Complete Reference, third ed., Springer, New
York, 2004.
[2]Lewis AS, Knowles G. Image compression
using the 2-D wavelet transform. IEEE
Transactions on Image Processing .1992.
[3]B. Ramakrishnan, N. Sriraam, Internet
transmission of DICOM images with effective
low bandwidth utilizationElsiver, 2006.
[4]C. Saravanan, M. Surender, Enhancing
Efficiency of Huffman Coding using
Lempel Ziv Coding for Image Compression,
International Journal of Soft Computing and
Engineering (IJSCE), Volume-2, Issue-6,
January 2013
[5]Victor Sanchez*, Student Member, IEEE,
Rafeef Abugharbieh, Senior Member, IEEE,
and Panos Nasiopoulos, Member, IEEE 3-D
Scalable Medical Image Compression With
Optimized Volume of Interest Coding, IEEE
Transactions On Medical Imaging, VOL. 29,
NO. 10, OCTOBER 2010

All the medical images having three layers. if

gray image means it is luminance components.
The signal noise ratio should be calculated
using the formula
PSNR (dB) =10log10 [(2552/MSE)]
Where MSE=

1
xy

(1)

x 1 y 1

[ I ( i,j) K ( i,j) ]
i= 0 j=0

(2)

Where x and y is the width and height of the

image, I is the original data and K is the
compressed data respectively. From table
(1),increased compression ratio for the image 4
is
obtained.Using
compression
ratio,
representation of the reduced file size by the
compressing algorithm can be obtained.
CR=Uncompressed file size/compressed file
Size
(3)

ISBN 978-93-80609-17-1
219

International Conference on Recent Advances in Mechanical Engineering and Interdisciplinary Developments [ICRAMID - 2014]

[6] Gabriela Dudek *, PrzemysOaw Borys,

Zbigniew J. Grzywna Lossy dictionary-based
image compression method .Image and Vision
Computing 25 ,PP 883889. 2007.
[7] J. Z. Wang, "Wavelets and imaging
informatics: A review of the literature" J.
Biomed. Inf., Apr. 2001 vol. 34, no. 2, pp. 129141,.
[8] Grgic S, Grgic M, Zovko-Cihlar B.
Performance analysis of image compression
using wavelets. IEEE Transactions on
Industrial Electronics; 48:68295, 2002
[9] Wei Li, Zhen Peng Pang, Zhi Jie Liu
SPIHT algorithm combined with Huffman
encoding. Third International Symposium on
Intelligent Information Technology and
Security Informatics, 2010 IEEE.
[10] Antonini M, Barland M, Mathieu P,
Daubechies I. Image coding using the wavelet
transform. IEEE Transactions on Image
Processing 1992

ISBN 978-93-80609-17-1
220