International Journal of Engineering Research & Technology (IJERT)

ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

Detection Of Tumour Using Grey Level Co-Occurrence Matrix And

Lifting Based DWT With Radial Basis Function
P. Mohanaiah 1 , Dr. P. Sathyanarayana2
1
Professor, Dept. Of E.C.E, N.B.K.R.IST, Vidyanagar India
2
Professor, Dept. Of E.C.E, S.V.U College of Engineering, Tirupati, India

ABSTRACT I. INTRODUCTION.
Magnetic Resonance Imaging (MRI) is an Due to the advent of computer
advanced medical imaging technique technology, image processing techniques
providing rich information about the have become increasingly important in a
human soft tissue anatomy. In this paper, a wide variety of applications. This is
tumour detection technique, applied to an particularly true for medical imaging such
MR Image using Grey Level Co- as Ultrasonography, CT, MRI, and Nuclear
RT
occurrence Matrix (GLCM) and Lifting Medicine which can be used to assist
DWT (LDWT) is proposed. Radial Basis doctors in diagnosis, treatment, and
IJE

Function Network (RBFN) is used to research. These systems are talented in

differentiate between abnormal tissues and discovering the structure, the function and
the normal tissues. The Comparative the diseases that influence the human
analysis is carried out with the Feed anatomy. It is highly difficult task to
Forward Neural Network (FFNN) and the categorize tumour types though MRI
obtained results are analysed in terms of seems to be competent in offering
the performance metrics. Simulation information considering the position and
results prove that the proposed technique volume of tumours. To overcome this,
have higher performance in terms of Biopsy and Spinal Tap methods are used,
Sensitivity Specificity and Accuracy. which are complex and time consuming
Keywords -- Grey Level Co-occurrence techniques. Various methods proposed in
Matrix, Lifting Discrete Wavelet the literature have met with only limited
Transform, Feed Forward Neural success due to overlapping intensity
Network, Radial Basis Function Network distributions of healthy tissues, tumour,
and surrounding edema. To overcome the
challenges in diagnosis, a tumour detection

IJERTV2IS60700 www.ijert.org 1677

 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

technique using GLCM and Lifting based Angular Second Moment (ASM), Entropy,
DWT is proposed. The proposed technique Inverse Difference Moment (IDM),
show higher performance in terms of Moment and Peak value of histogram are
Sensitivity, Specificity and Accuracy. the texture features computed.
Feature extraction using GLCM is Minimum: This minimum value can be
presented in section II, feature extraction used to differentiate the abnormal images
using LDWT is presented in section III, since most of the abnormal regions are
tumour detection using RBFN is given in pure white pixels.
section IV, and finally, simulation results Maximum: Other feature used in the
and conclusions are presented in sections proposed method is maximum value. From
V and VI respectively. the corresponding GLCM features image a
maximum value is selected.
II. Feature Extraction Using GLCM. The ASM is known as uniformity or
One of the most famous texture energy. It measures the homogeneity of an
study techniques, GLCM [1], assesses image. When pixels are very similar, the
image properties associated to Second- ASM value will be large.
RT
Order statistics. Using Gray Level Co- G 1 G 1
ASM  { p(i, j )}2 .......... (1)
occurrence Matrix (GLCM), the i 0 j 0
IJE

statistical characteristics from MR This measure of contrast or local intensity

images are obtained are also known as variation will favour contributions from
Gray Level Spatial Dependence Matrix P(i, j) away from the diagonal, i.e. ‘i’≠’j’.
(GLSDM). Haralick[2] put forward Entropy refers to the quantity of energy
GLCM, which is a statistical strategy that is permanently lost to heat every time
that can well explain the spatial a reaction or a physical transformation
connection among pixels of dissimilar gray occurs. Entropy cannot be recovered to do
levels. GLCM is a two dimensional useful work. Because of this, the term can
histogram in which (i, j) the element is the be understood as amount of irremediable
frequency of event ‘i’ that takes place with chaos or disorder. The equation of entropy
‘j’. Texture feature extraction is is:
categorised into two steps: the first step is G 1 G 1
Entropy   p(i, j ) log p(i, j )
computing the co-occurrence matrix and i 0 j 0
....... (2)
the second step is calculating texture IDM is usually called homogeneity that
features based on the co-occurrence matrix measures the local homogeneity of an
[3][4]. Minimum value, Maximum value,

IJERTV2IS60700 www.ijert.org 1678

 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

image. IDM feature obtains the measures Update : sk(r)= sk(r-1) + uj(r) dk+j(r) ... (7)
of the closeness of the distribution of the Normalize:sk(R)=K0sk(R),dk(R)=K1dk(R) ...
(8)
GLCM elements to the GLCM diagonal.
G 1 G 1
1
IDM   p(i, j ) ....... (3)
j 0 1  (i  j )
2
i 0

Similarly moment and Peak value of

histogram is computed. The block diagram
of the feature extraction using GLCM
process is shown in figure 2.1.

Fig.2.2. Lifting scheme block diagram

Lifting based DWT requires less

computation time, less memory [6][7], and
easily produces integer-to-integer wavelet
transforms for lossless compression.
RT
Similarly, Second Order statistical features
are extracted from the Lifting DWT MR
IJE

Image as shown in figure 3.1.

Fig.2.1. Feature extraction using GLCM.

The obtained feature for a given MR

image is applied to a RBFN.

III. Lifting DWT.

The lifting scheme is an alternative method
of computing the wavelet coefficients [5].
A spatial domain construction of bi-
Fig.3.1. Feature extraction using LDWT.
orthogonal wavelets consists of the
following four basic operations as shown
The features extracted are then applied to
in figure 2.2. These are summarized as
RBFN.
follows.

Split : sk(0)=x2i(0), dk(0)=x2i+1(0) …(5)

Predict : dk(r)= dk(r-1) – pj(r) sk+j(r-1) …(6)

IJERTV2IS60700 www.ijert.org 1679

 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

IV. Tumour detection using RBFN. Based on the computed TP, FP, FN, and
The features extracted using GLCM TN values the evaluation metrics can be
method and LDWT are applied to a RBF defined as follows
Network. RBFN consists of three layers, Sensitivity is the fraction of true positives
namely an input layer, hidden layer and that are correctly identified by a diagnostic
out layer. The most popular Gaussian test. It shows how good the test is at
Activation Function is used in Radial detecting a disease.
Basis Function. Classification of MRI Sensitivit y  TP/(TP  FN)
images using RBFN requires training and Specificity is the fraction of the true
testing phase. In training phase the negatives correctly identified by a
network is trained with around 30 tumour diagnostic test. It suggests how good the
as well as non tumour images. In the test is at identifying normal (negative)
testing phase the candidate MR Image is condition.
applied to the system. The performance of Specificity  TN/(TN  FP)
the system is evaluated using three Accuracy is the fraction of true results,
parameters Sensitivity, Specificity and
RT
either true positive or true negative, in a
Accuracy [9]. These values are computed population. It measures the degree of
from the parameters True Positive (TP),
IJE

veracity of a diagnostic test on a condition.

True Negative (TN), False Negative (FN)
and False Positive (FP) for each of the Accuracy  (TN  TP)/(TN  TP  FN  FP)
input image, based on the definitions given
in table I. V. Simulation Results.
Table I: Defining the terms TP, FP, FN,
and TN. The MRI image dataset utilized here is
Condition as taken from the publicly available sources.
determined by the GLCM and LDWT applied on the input
Outcome Row Total
Standard of Truth
MR Image (A) are shown in (B) and (C)
Positive Negative
respectively in figure 2.4.The features
Positive TP FP TP+FP
Negative FN TN FN + TN computed for a given MR image is
Column tabulated in table II.
TP+FN FP+TN N= TP+TN+FP+FN
total The obtained results obtained with the
proposed technique are compared with
FFNN [8] based technique. The
comparative analysis is given in table III.

IJERTV2IS60700 www.ijert.org 1680

 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

Table III: Performance Metrics.

Parameters FFNN RBFN

Sensitivity 0.94 0.91

Specificity 0.65 1
Accuracy 0.79 0.93

From table III, it is evident that the

Accuracy of the technique is observed as
93%.The Specificity of the technique is
also proved to be maximum. But the
Sensitivity of the proposed technique is
slightly less than FFNN based system.
This method is a simple method using
which tumour can be detected easily.
RT
VI .Conclusions & Future Scope.
Tumour detection technique using both
IJE

GLCM features and Lifting DWT features

applied to a RBF network is proved to be a
successful technique for the detection of
Fig 4.1. (A) Input MR Image, tumour in MR Images as compared with
(B)GLCM applied Image., (C) LDWT FFNN.
applied Image The future scope of the work deals in
Table II: Features extracted with values improving the sensitivity and accuracy in
Features GLCM LDWT detecting Tumours of MR Images using
Min 0 -607 this technique.
Max 70811 26066

ASM 7319650424 368370697

Entropy 0.928362072 0.938712075

IDM 2260.057904 0.184747626

Moment 4399360510392.18 42676.8488

Peak of histogram 42 96052

IJERTV2IS60700 www.ijert.org 1681

 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

References. Implementation of a Lifting Scheme 2D

[1] Mari Partio, Bogdan Cramariuc, DWT Architecture", International Journal
Moncef Gabbouj and Ari Visa, "Rock of Recent Technology and Engineering,
texture retrieval using gray level co- vol. 2, no.1, p p. 34-38, March 2013.
occurrence matrix", Journal of Theoretical [7] A. F. Mulla, Dr.R. S. Patil, "Lifting
and Applied Information Technology, Based 2D DWT Processor for Image
vol.33. no.2, p p. 155-164, 2011. Compression”, International Journal of
[2] R.E. Haralick, K.Shanmugam, Engineering Science and Innovative
I.Dinstein, Textural Features for Image Technology (IJESIT) Volume 1, Issue 1,
Classification, IEEE Transactionson September 2012.
Systems, Man and Cybernetics, Vol. [8] Mohaniah. P. Sathyanarayana, A. S.
SMC-3, No.6, Nov 1973 Ram Kumar Reddy and A. Vijayalakshmi,
[3] R.S.RajKumar and G.Niranjana, "Co- "Design and Implementation of 3-D DWT
occurrence matrix and its statistical for Video Processing Applications",
features as a new approach for face International Journal of Advanced
recognition" International Journal of Electrical and Electronics Engineering,
RT
Research in Engineering & Advanced vol.1, no.3, p p. 45-50, 2012.
Technology, vol. 1, no. 1, p p. 1-7, March [9]Arati Kothari, "Detection and
IJE

2013. classification of brain cancer using

[4] P. Mohanaiah, P. Sathyanarayana and artificial neural network in MRI images",
L. GuruKumar, "Image texture feature World Journal of Science and Technology,
extraction using glcm approach" vol. 2, no. 5, p p. 1-4, 2012.
International Journal of scientific & [10]Pauline John, "Brain Tumour
Research Publications,Conference on Classification Using Wavelet and Texture
Communications & Signal Processing, vol. Based Neural Network", International
3.Issue no. 5,ISSN no:2250-3153 May Journal of Scientific & Engineering
2013. Research, vol. 3, no. 10, p p. 1-7, October
[5] Nitish Zulpe and Vrushsen Pawar, 2012.
GLCM "Textural Features for Brain
Tumor Classification”, International
Journal of Computer Science Issues, Vol.
9, Issue 3, No 3,p.p 354-359, May 2012.
[6] Naseer M. Basheer and Mustafa
Mushtak Mohammed, "Design and FPGA

IJERTV2IS60700 www.ijert.org 1682

 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 2 Issue 6, June - 2013

AUTHORS: visited many countries like Malaysia,

Mohanaiah P – U.S.A and other countries as visiting
has received B.Tech faculty. He is the principal investigator for
(Electronics & DST sponsored projects. He is the member
Communication of expert committee of NBA and
Engineering ,1992), chairman, BOS (E.C.E) for P.G. in S.V.
M.Tech (Atmospheric Science & University, Tirupati and V.S. University,
Engineering, 1996) and also Pursuing Nellore, A.P., India. He presented 16
Ph.D. (Signal Processing) from S.V. papers in International Journals like IEEE
University, Tirupati, A.P, India. Presently & IETE. He also Presented more than 30
he is working as Professor and Head, Dept. papers in NationalConferences.
of ECE, N.B.K.R.I.S.T, Nellore, A.P.,
India. He has presented papers in two
International conferences and five National
conferences in various fields, published
two international journals. He is a life
RT
member of ISTE, and member of IEEE.
His Areas of Interest also include Video &
IJE

Image Processing, Speech Processing &

Radar Signal Processing.,

Dr. P. Sathyanarayana
– has received B.Tech
(E.C.E, 1976), M.Tech
(Instrumentation &
Control systems, 1978)
and Ph.D. (Digital Signal Processing,
1987) from S.V. University Tirupati. He
also obtained his Post Doctoral Fellow in
1-D Signal Processing from Concordia
University, Montreal, Canada from 1988-
1990. He worked as Head, Dept. of E.C.E,
S.V. University, Tirupati, A.P., India. He

IJERTV2IS60700 www.ijert.org 1683