Performance analysis of different feature-classifier binomials in motor-imagering BCIs: Preliminary results

Brain Computer Interface Systems (BCIs) allow the identification of volitive brain activity patterns. This allows their use as input channels for alternative communication and computer access systems by patients suffering from severe motor disabilities. This paper presents preliminary results obtained after extracting four different features from EEG signals in order to recognize the activity patterns by means of four different classifiers. The final goal is to determine the proper "feature - classifier” binomial for each user in order to increase system reliability and satisfaction in use.