Motor imagery EEG classification using feedforward neural network

Electroencephalography (EEG) is a complex voltage signal of the brain and its correct interpretation requires years training. Modern machine- learning methods help us to extract information from EEG recordings therefore several brain-computer interface (BCI) systems use them in clinical applications. By processing publicly available PhysioNet dataset, we extracted that could be used for training feedforward neural network classify three types activities performed by 109 volunteers. While volunteers were performing different activities, BCI2000 system was recording their signals 64 electrodes. We motor imagery runs where target appeared on either top or bottom screen. The subject instructed imagine opening closing both his/her fists (if top) feet bottom) until disappears EEGLAB Matlab toolbox applied feature extraction techniques. Then evaluated classification performance feedforward, multilayer perceptron (MLP) networks with structures (number layers, number neurons). Achieved accuracy score test data 71.5%.