An Effective Detection Method o f Glioma Tissue in Multimodal Brain Image u sing Adaptive Segmentation Algorithm
Perumal.B1, Sindhiya Devi.R2, Pallikonda Rajasekaran.M3

1B.Perumal, Associate Professor/ECE, Kalasalingam Academy of Research and Education, Srivilliputhur, India.
2R.Sindhiya Devi, Research Scholar/ECE, Kalasalingam Academy of Research and Education, Srivilliputhur, India.
3M.Pallikonda Rajasekaran, Professor/ECE, Kalasalingam Academy of Research and Education, Srivilliputhur, India. 

Manuscript received on 10 August 2019. | Revised Manuscript received on 17 August 2019. | Manuscript published on 30 September 2019. | PP: 7082-7090 | Volume-8 Issue-3 September 2019 | Retrieval Number: C5761098319/2019©BEIESP | DOI: 10.35940/ijrte.C5761.098319
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: In the medical field, accurate and faster classification and segmentation of tumor images in the brain is enormously crucial. In our paper, we introduce a distinct process to segment the multimodal brain image and to find out the tissues affected by glioma. Skull stripping is the preprocessing method to remove non-cerebral tissues. Later, the image is segmented using an Adaptive segmentation Algorithm (ASA) process, after doing primary segmentation to find out the consistent regions for the process of integrating images. It reduces over segmentation and conserves the frontiers of the image. The adaptive segmentation process follows the four steps: exemplification of the region and similarity measure, marking of the object and background, merging rule based on maximal homogeneity and the process of unifying. The segmented output image can clearly view the region that selected. From the image segmentation, several features are extracted including mean, variance, entropy using Gaussian Inception (GI) method. After feature extraction, classification is done using Support Vector Machine (SVM) classifier which classifies the voxels into normal and affected one. This proposed method could be used in the easier and effective diagnosis of brain tumour in brain MRI images. The diagnosis of brain tumour by using this interactive adaptive segmentation will be more accurate as in human investigation and at the same time classification is done at a faster rate as in fully computerized investigation. The method is evaluated experimentally using multimodal brain images of patients with glioma. The classification results of the proposed methodology are compared with the existing method which shows the improved accuracy.
Keywords: Glioma, Merging Process, SVM Classification, Gaussian Inception.

Scope of the Article: Classification