Mental Arithmetic Task Classification using Fourier Decomposition Method
Binish Fatimah, Abhishek Javali, Haaris Ansar, B G Harshitha, Hemant Kumar
Abstract
Solving an arithmetic problem is a complex task which involves fact retrieval, memory, sequencing and decision making. Automatic detection of such an activity from EEG signals will help in understanding of brain response to these cognitive tasks. In this work, we propose a mental arithmetic task detection algorithm from a single lead EEG signal. Fourier Decomposition method is used to decompose the signal into M uniform sub-bands and features, like energy, entropy, and variance, are computed from each of these sub-bands. Kruskal-Wallis method has been used to select only the statistically relevant features. These selected features are, then, used to classify the given EEG dataset into two classes using support vector machine with cubic kernel. To validate the efficacy of the proposed algorithm, simulation results are presented using dataset available on MIT PhysioNet, titled EEG during mental arithmetic task.