Proposing an Efficient Method to Classify MRI Images Based on Data Mining Techniques

Nowadays, Magnetic Resonance Images (MRI) is the most common tool for diagnosis of soft tissues. Using fully automated classification magnetic resonance images of the human brain that are important for clinical research studies, can be detect the healthy or sick person. This paper purpose enhances the classification accuracy and achieves good performance in classifying the MRI images. Automatic classification of images uses the models and formal criteria involving key stages of feature extraction, feature reduction and learning algorithms. In this study, the best-known and most effective feature extraction algorithms, reducing the features, such as wavelet transform principal component analysis. Also a hybrid approach to improve the efficiency of the support vector classifier using Cuckoo evolutionary algorithm has been employed. While the initial values in support vector algorithm are set manually, the values in proposed algorithm retrieved from Cuckoo algorithm automatically. The proposed method is evaluated using these criteria: accuracy, sensitivity and specificity. The results of the combined classification accuracy, sensitivity and specificity are above 98% that stronger and more effective compared with other recent work