Extraction of Wavelet Based Features for Classification of T2-Weighted MRI Brain Images

Extraction of discriminate features is very important task in classification algorithms. This paper presents technique for extraction cosine modulated feature for classification of the T2-weighted MRI images of human brain. Better discrimination and low design implementation complexity of the cosine-modulated wavelets has been effectively utilized to give better features and more accurate classification results. The proposed technique consists of two stages, namely, feature extraction, and classification. In the first stage, the energy features from MRI images are obtained from sub-band images obtained after decomposition using cosine modulated wavelet transform. In the classification stage, Mahalanobis distance metric is used to classify the image as normal or abnormal. Average Classification accuracy with a success rate of 100% has been obtained.