RESEARCH ARTICLE
Segmentation of glioma tumors using convolutional neural
networks
R. Anitha
1
|
D. Siva Sundhara Raja
2
1
Faculty of Electronics and
Communication Engineering, Adhi
College of Engineering and Technology,
Chennai, Tamilnadu, India
2
Faculty of Electronics and
Communication Engineering, SACS
MAVMM Engineering College, Madurai,
Tamilnadu, India
Correspondence
R. Anitha, Faculty of Electronics and
Communication Engineering, Adhi
College of Engineering and Technology,
Chennai, Tamilnadu, India.
Email: anithagodavarthy@gmail.com
Abstract
The abnormal development of cells in brain leads to the formation of tumors in brain.
In this article, image fusion based brain tumor detection and segmentation methodol-
ogy is proposed using convolutional neural networks (CNN). This proposed
methodology con sists of image fusion, f eature extraction, cla ssific ation, and segmenta-
tion. Discrete wavelet transform (DWT) is used for image fusion and enhanced brain
image is obtained by fusing the coefficien ts of th e DWT transform. Further , Grey
Level Co-occurrence Matrix features are extracted and fed to the CNN classifier for
glioma image classificatio ns. Then, morphologi cal operations with closing and open-
ing functions are used to segment the tumor region in classified glioma brain image.
KEYWORDS
brain tumors, classifier, features, glioma, image fusion
1
|
INTRODUCTION
As per the report 2016 of American Brain Tumor Association
(ABTA), 61 200 persons in United States of America (USA)
are affected by brain tumor. WHO (2016) reported that the
persons affected by brain tumor is 6.2 million worldwide. The
tumor is formed in brain due to the abrupt development of the
cells in brain region. The initial stage of the tumor formation
in brain region is primary brain tumors (PBT) and they are
further treated as benign. The tumors in this category are hav-
ing homogeneous structure s. The severe stage of the tumor in
brain region is secondary brain tumors (SBT) which is also
called as malignant. The brain tumors are differentiated with
their sha pe, size, intensity, and d ensit y. The benign tumo rs
can be controlled and cured by medication. The untreatable
benign tumors become malignant tumors. The malignant
tumors are the uncontrollable tumors which can be cured only
by surgery. The brain tissues in brain are gray matter, white
matter, and cerebro spinal fluid. These brain tissu es play an
important role in the process of brain tumor classification. The
tumors are having different shape, size, and located in differ-
ent areas in brain region which makes the tumor detection and
segmentation process more complex. The brain tumor seg-
mentation by physic ian is tim e consumi ng and err or probe
process. The manual detection of brain tumor is not suitable
for large population countries. Hence, there is a need for auto-
matic detection and segmentation of brain tumors. The malig-
nant tumors in brain image are classified as glioma and
meningioma. Figure 1A shows the glioma tumor brain mag-
netic resonance imagi ng (MRI) and Figu re 1B shows the
meningioma tumor affected brain image.
2
|
LITERATURE SURVEY
Bahadure et al.
1
used wavelet transform method to detect
abnormal lesions in brain region. The authors used Berkeley
type wavelet kernels which were symmetric with its decom-
position level. This multi level decomposition sub-bands
were trained properly with support vector machine (SVM)
classifier. The sensitivity rate about 97.72%, specificity rate
about 94.2%, and accuracy rate about 96.51% were achieved
by authors on BRATS 2015 dataset. Sreedhanya and Pawar
2
applied hybrid classification technique on preprocessed brain
MRI image. Gaussian mixture model was developed by
authors to represent the texture features of the brain image
for improving the segmentation accuracy. Fast Fourier trans-
form was used by Alfonse and Salem
3
to transform the
spatial domain features into frequency domain features.
Maximal-relevance classifier was used to train the extracted
354
|
V
C
2017 Wiley Periodicals, Inc. wileyonlinelibrary.com/journal/ima Int J Imaging Syst Technol. 2017;27:354–360.
Received: 6 July 2017
|
Revised: 25 July 2017
|
Accepted: 16 August 2017
DOI: 10.1002/ima.22238