Brain tumor detection and classification with feature extraction and reduction using DWT and PCA

Abstract

The central control unit of human body is brain. The tumor is not diagnosed in early stage then it affects the brain means it causes the death of the patient.Magnetic Resonance Image (MRI) doesn’t produce any harmful radiation and it is a better method for area calculation as well as classification based on the grade of the tumor. Nowadays there exists no automatic system to detect and identify the grade of the tumor. This paper proposes brain tumor classification which is divided into four phases as pre-processing, segmentation, feature reduction and extraction, classification. Segmentation of brain Tumor is a one of the basic steps in detection and classification of tumor. Median filter is used to eliminate the noise and Combination of K means cluster and otsu binarization is used to segment the brain tumour. DWT (Discrete wavelet transform) and GLCM (Grey Level co-occurrence matrix) used for transform and spatial feature extraction and PCA (Principal component analysis) reduces the feature vector to maintain the classification accuracy of brain MRI images. For the performance of MRIs classification, the significant features have been submitted to KSVM (kernel support vector machine).The proposed system will reduce processing time and better accuracy can be achieved. The proposed method is validated on BRATS 2015 dataset.