COGNITIVE ASSESSMENT THROUGH THE ANALYSIS OF EEG SIGNALS

Abstract

With the advent of emerging technologies and methodologies of biomedical signal processing, research on the cognitive sciences has become one of the most innovative, enthralling and challenging researches in the field of biomedical engineering. The problem of cognitive assessment and enhancement has gained major importance amongst the today’s cognitive researches, as it aids in identification and treatment of cognitive related disorders like Attention Deficit Hyperactivity Disorder (ADHD), Spatial Navigation, Short Term Memory, Cross Modal Processing etc. In this work, as a part of Cognitive Assessment, we are concentrated towards the testing of Working Memory and Cognitive Workload through the analysis of the two channel ( - ) EEG signals obtained from the subjects when they were presented with some familiar stimulus. For this, the subjects were explained about the stimuli related to a situation and later were shown these stimuli and a model had been developed to differentiate these different types of stimuli responses. Several features in conjunction with classifiers have been explored and the corresponding results were analysed to decide the optimum feature and classifier that can be selected for obtaining the optimum classification accuracy. A maximum classification accuracy of 66.67% had been obtained when tried with the LDA+RBFFNN classifier for -channel when the feature of Hurst Exponent on 3-5 IMFs is used for the task of inter Stimuli Classification and a maximum classification accuracy of 100% had been obtained when tried with the LDA+KNN classifier for and channels when the feature of Hurst Exponent is used for the task of inter Subject Classification and the results show a very good overall Inter subject classification accuracies for the classifier LDA+KNN.