Signaleeg : A practical tool for EEG signal data mining

Abstract

Due to the proliferation of brain and neurological disorders (World Health Organization 2006), EEG (Blinowska and Durka 2006) is gaining attention as a support for decision making in the fields of neurology, psychology, and psychiatry. But EEG data are not always easy to understand. Therefore, extracting the desired information from EEG data in different contexts is an important requirement. This article analyses state-of-the-art EEG signal processing tools and proposes a new one: Signaleeg. This addresses the limitations of previous tools. It has been designed with the aim of helping users to build predictive models from EEG signals in a process that is called signal-data mining (DM). Moreover, Signaleeg is user friendly and multi-threaded, with optimisation facilities for finding the best predictive model. It has been implemented and tested in three scenarios: schizophrenia diagnosis, alcoholism detection, and emotion recognition. The tool provided good results in each case, thus demonstrating its versatility.

Keywords: Alcoholism; EEG; Emotions; Schizophrenia; Signal characterization; Toolbox.