Spectral regression discriminant analysis for brain MRI classification

In this paper, a new method for brain magnetic resonance imaging (MRI) classification based on spectral regression is proposed. In feature extraction step, the primary features are obtained using a three-level two-dimensional discrete wavelet transform (2D DWT). The dimension of primary feature vector is high and classifying such high-dimensional vector requires huge computational complexity. We propose to use spectral regression discriminant analysis (SRDA) to reduce the dimension of the feature vector. Then, support vector machine (SVM) is used to classify low-dimension feature vector. We consider ten-class brain disease problem and evaluate the performance. The results indicate that the proposed approach can determine the type of brain MRI disease with high accuracy, and outperforms recently presented algorithms and it has less computational complexity.