Cortical activities pattern recognition for the limbs motor action

Unlocking the neural code of the human´s brain has long been the main focus of neuroscience studies with intensive research carried out to model the brain. Some of these techniques use the magnetic resonance imaging (MRI) and electroencephalographic (EEG) systems. The EEG signals estimate the cortical activities using non-invasive Brain Computer Interfaces (BCI) with scalp potential measurements targeted at areas of interest according to the Brodmann areas. In this paper, The EEG signals around the primary motor cortex and the Cerebellum are believed to be responsible for the motor action of the human limbs. Based on these early findings that shows correlation between the EEG signals and the corresponding motor action has allow detail investigation for best feature extraction technique. Neural Network and fuzzy neural network algorithms were then used to classify the different motor actions. Two feature extraction techniques were explored based on the Gaussian mixture model (GMM) and the Mel frequency cepstral coefficients (MFCC). Multi layered perceptron (MLP), radial basis function (RBF), the adaptive network-based FIS (ANFIS) and the generic self-organizing fuzzy neural network (GenSoFNN) were used as classification tools. Results of the experiments indicate the potential of using MFCC for feature extraction in recognising the motor activities from the brain.