Clustering of EEG occipital signals using k-means

Recent studies show that it is feasible to use electrical signals from Electro-encephalography (EEG) to control devices or prostheses, these signals are provided by the body and can be measured on the scalp to determine the intent of the person when it is observing a visual stimulus frequency range detectable by the human eye. This group of signals are very susceptible to noise due to voltage levels that are able to acquire. Therefore, in this work we propose a statistical analysis of the distribution of normal EEG signals in order to determine the need of a pre-processing to remove noise components from electrical grids or other possible sources. This preprocessing includes the design and use of a filter that will eliminate any signal component that is not in the operating frequency range of the EEG occipital area of the brain. Finally, we will proceed to use the k-means algorithm to cluster with signals according to their frequency and temporal characteristics.