Classification of Brain MRI Based on LH and HL Sub-bands of Wavelet Transform

Automated classification of brain MRI is important for the analysis of tumor. In this paper magnetic resonance imaging (MRI) is considered to solve the problem of automatic classification of brain images. It presents the classification system with three stages. It consists of discrete wavelet decomposition of the image, texture feature extraction from the LH and HL sub bands and final classification by probabilistic neural network (PNN).It shows that the horizontal (LH) and vertical (HL) sub bands of the wavelet transform gives higher performance compared to LL sub band. It can effectively encode the discriminating features of normal and abnormal images. Keywords-MRI, DWT, texture feature, PNN