Abnormality detection and classification from brain MRI using machine learning

In our proposed method, wavelet transform is inked first to excerpt features from MR image, and then to lessen the dimensions of features the principle component analysis has been applied. The lessen features have been given as input to a Support Vector Machine (SVM) that uses kernel function for transforming the lower dimension feature space to upper dimension feature space. An approach of K-fold-stratified-cross validation has been taken care of to improve the generalization of kernel support vector machines. The proposed method has used different kernels and observed that the Gaussian Radial Basis function represents the accuracy result as 90.3% which is the highest.