Functional Magnetic Resonance Imaging (fMRI) considers as a powerful technique for measuring and mapping the brain activities. The significant advantages of fMRI are non-invasive process and no use of the radiations. It evaluates the brain function safely and effectively. The fMRI helps to ensure the variations in brain functions such as functional disorders, network in cognitive tasks, effects of medication therapies, and so on. The complex behaviour of fMRI data analysis needs efficient techniques for improving the brain activity classification and prediction. In this paper, an efficient cognitive state classification method proposed namely Genetic Algorithms (GA) and Support Vector Machine (SVM). The GA method chooses features that are most relevant in minimum time and forward it to the SVM classifier. The SVM removes the irrelevant voxels and calculate the informative spatial patterns. In the experimental analysis, the proposed approach classifies either the person is viewing image or reading the sentence through recognizing the object. The proposed method performance measured in terms of sensitivity, specificity, computation time and accuracy. An experimental outcome showed that the proposed methodology delivered an average accuracy of 98.9 % in cognitive state classification.
Volume 12 | 01-Special Issue
Pages: 46-54
DOI: 10.5373/JARDCS/V12SP1/20201045