An improved brain tumor segmentation and classification method using SVM with various kernels

Abstract

The objective of this paper is to propose an efficient brain tumor segmentation and classification approach using SVM. In
the first phase, the tumor segmentation is carried out using smoothing, skull stripping, filtering, enhancement and
identifying the region of interest. The resultant tumor is classified in the second phase based on its characteristic features.
The features are extracted using Discrete Wavelet features, and Grey Level Co-occurrence Matrix and selected using
Principle Component Analysis. Finally, the tumor categories are classified by Support Vector Machine using various
Kernels. Among the various Kernels, Gaussian Radial Basis Function Kernel resulted in better performance.