Fusion based Glioma brain tumor detection and segmentation using ANFIS classification
Introduction
The morphology behavior of the cells in brain is affected by improper mitosis process. This leads to the formation of tumor cells in brain region. These tumor cells are having different morphological properties such as size and intensity. Most of the tumor cells in brain region are low contrast with respect to the other surrounding cells. These abnormalities in brain image are identified by scanning the brain regions using Magnetic Resonance Imaging (MRI) technique [10]. The detection and segmentation of intratumor region in brain MRI image is a challenging task due to the low intensity variation between tumor cells and its surrounding cells in brain image. In current medical scanning methods, MRI scanning procedure is superior to Computer Tomography (CT) due to its high sensitivity and high contrast with respect to various intensity tissues in brain image. The brain tumors are categorized into Glioma and Glioblastoma [9]. The Glioma tumors are high pixel intensity cells and irregular boundary regions (Bandhyopadhyay et al. [3]). Glioblastoma tumors are low pixel intensity cells and it can be detected by many conventional methods with high level of accuracy. The detection and segmentation of Glioma brain tumors in brain MRI image is a challenging task due to its irregular boundary regions (Funmilola et al. [7]). In common, the Glioma tumor images are categorized into low grade Glioma tumors and high-grade Glioma tumors based on its severity level. In this paper, ANFIS classification approach-based Glioma brain tumor detection and segmentation methodology is proposed in an automated manner. The main purpose of this paper is to develop an efficient system which localizes the tumor boundary with high level of accuracy. Fig. 1 shows the Glioma brain MRI image which clearly represents the irregular boundary region of tumor cells.
This paper is structured as, Section 2 states various conventional methodologies for Glioma brain tumor detection, Section 3 proposes an efficient methodology for brain tumor detection and segmentation using ANFIS classification approach, Section 4 discusses the simulation results of the proposed Glioma tumor segmentation method with respect to other state-of-arts methods and Section 5 concludes the paper by stating its advantages and future developments.
Section snippets
Literature Survey
Anitha et al. [2] differentiated the normal brain MRI image from abnormal MRI brain image using Convolutional Neural Network (CNN) approach. The authors used Maxpool methodology in CNN architecture in order to improve the classification accuracy of the brain tumor detection system. The authors achieved 88.8% of sensitivity, 91.6% of specificity and 92.1% of accuracy on Leader Board data subset of BRATS dataset. The authors achieved 91.2% of sensitivity, 93.4% of specificity and 93.3% of
Methods
This paper proposes an image fusion based Glioma brain tumor detection and segmentation methodology using ANFIS classification approach. Fig. 2 shows the proposed brain MR image fusion using NSCT transform coefficients. It fuses low frequency and high frequency coefficients and inverse NSCT transform is applied over these fused coefficients in order to obtain fused brain MR image.
Results and Discussions
In this paper, the proposed brain tumor detection methodology is applied on the MRI brain images which are accessed from publicly open access dataset BRATS 201516. The proposed algorithm is simulated using MATLAB R2014b version with 3.1 GHz GPU Processor and 4GB internal RAM as hardware devices. This open access BRATS 2015 dataset [4] consists of three different brain MRI image sub datasets as Training, Leaderboard and Challenge. The training sub dataset consists of 20 High Grade Tumor (HGT)
Conclusions
This paper proposes a methodology to detect and segment the Glioma tumors in brain MR image. The method uses fusion technique based on NSCT transform. The enhanced image by fusion technique is applied to the feature extraction process. The extracted texture features are classified using ANFIS classifier. The proposed methodology is applied on both low grade and high grade Glioma tumor MR images in BRATS open access dataset. The results obtained from the proposed methodology are compared with
Conflict of Interest
There is no conflict of Interest in this paper
References (14)
- et al.
Image processing techniques for automatic detection of tumor in human brain using SVM
Int. J. Adv. Res. Comput. Commun. Eng.
(2015) - et al.
Segmentation of glioma tumors using convolutional neural networks
Int. J. Imaging Syst. Technol.
(2017) - Bandhyopadhyay, S.K, Paul, T.U. Automatic segmentation of brain tumour from multiple images of brain MRI, Int. J. Appl....
- et al.
Tumor segmentation in brain MRI using a fuzzy approach with class center priors
EURASIP J. Image Video Process.
(2014) - et al.
Brain tumor detection using artificial neural networks
J. Sci. Technol.
(2012) - et al.
Fuzzy K-C-means clustering algorithm for medical image segmentation
J. Informat. Eng. Appl.
(2012)
Cited by (115)
Preparation of an aptamer electrochemical sensor for the highly sensitive detection of glioma cells
2023, International Journal of Electrochemical ScienceBRAIN TUMOR SEGMENTATION AND CLASSIFICATION USING OPTIMIZED U-NET
2024, Journal of Mechanics in Medicine and BiologyImage Fusion Techniques and Applications for Remote Sensing and Medical Images
2024, Deep Learning for Multimedia Processing Applications: Volume One: Image Security and Intelligent Systems for Multimedia ProcessingA Deep Learning Based Glioma Tumour Detection Using Efficient Visual Geometry Group Convolutional Neural Networks Architecture
2024, Brazilian Archives of Biology and Technology