An Efficient Brain Tumor Classification Based on SOBS Method for MRI Brain Images
M. Mohamed Sathik1, E. Synthiya Judith Gnanaselvi2

1Dr. M. Mohamed Sathik*, Principal, Sadakathullah Appa College, Tirunelveli, India.
2E. Synthiya Judith Gnanaselvi, Research Scholar, Bharathiar University, Coimbatore, India
Manuscript received on September 16, 2019. | Revised Manuscript received on October 15, 2019. | Manuscript published on October 30, 2019. | PP: 826-833 | Volume-9 Issue-1, October 2019 | Retrieval Number: A9379109119/2019©BEIESP | DOI: 10.35940/ijeat.A9379.109119
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: In the field of medical imaging, the segmentation and classification of brain tumors is a complex and important area of studies because it is essential for the intention of early tumor diagnosing and treatment of brain tumors and other neurologic complaints. Earlier segmentation methods require huge number of iterations, longer time and a reduced accuracy. Therefore, this article proposes a multi-stage strategy whereby tumor segmentation and classification can be accurately performed with lower error rate. The proposed system incorporates three phases such as prediction, segmentation along with morphological operations to solve the discontinuities. The proposed segmentation method is named as Self Organisation Based Segmentation (SOBS) method. It is compared with some of the deformable models in literature. Next Use the Gray Level Co occurrence Matrix to extract features. Finally Use the Gray Level Co-occurrence Matrix to extract features and classify them into normal or abnormal. If it is classified as abnormal, then again classify into glioma or meningioma. The performance metrics such as accuracy, PSNR and MSE are used for scrutinize the performance of these methods. From the investigational outcomes, the classification accuracy was found to be very high using the proposed segmentation method SOBS with the Random Forest (RF) Classifier.
Keywords: Random Forest, Radial Basis Function, Segmentation, Classification, Self Organisation Based Segmentation (SOBS).