Analysis of EEG Mapping Images to Differentiate Mental Tasks in Brain-Computer Interfaces - Foundations on Natural and Artificial Computation - page 245

Analysis of EEG Mapping Images to
Differentiate Mental Tasks in Brain-Computer
Interfaces

Andres Ubeda, Eduardo Ianez, Jos´eM.Azor ın, and Eduardo Fernandez

Biomedical Neuroengineering Group, Miguel Hernandez University of Elche
Av. de la Universidad S/N, 03202, Elche, Spain
{ aubeda,eianez,jm.azorin,e.fernandez } @umh.es
http://nbio.umh.es

Abstract. This paper describes a study of a new classifier based on
EEG mapping analysis in order to develop a Brain-Computer Interface
(BCI) through computer vision techniques. To this end, the data from
three different subjects (BCI Competition Data Set V) have been studied
to show proper EEG maps of the three mental tasks registered. A new
classifier based on image analysis of the EEG maps has been presented
as a suitable way to distinguish between the different tasks, showing
in which conditions of frequency and time the images obtained for each
mental task can be best classified. The classifier has been tested obtaining
the success percentage of classification of each subject showing that this
kind of techniques are able to classify between three mental tasks with
good results.

Keywords: Brain-Computer Interface (BCI), Electroencephalography
(EEG), EEG Mapping, Computer Vision.

1 Introduction

A Brain-Computer Interface (BCI) is based on the processing of brain signals
in order to generate commands to control external devices [1]. Brain signals
can be registered using invasive or non-invasive methods. Invasive methods are
based on the using of microelectrodes implanted directly in the brain. These
techniques have been used in animals to determinate the movement intention [2]
or to control a cursor in a screen [3]. In humans, the use of invasive techniques
has ethical implications as well as medical risks. For this reason, electrodes are
placed on the scalp of the patient to obtain the electroencephalographic (EEG)
signals [4].
Non-invasive BCIs are commonly used in different applications such as the
control of a robot [5,6] or the keyboard in a computer [7,8]. Non-invasive BCIs can
be classified as evoked and spontaneous. In evoked interfaces, the EEG signals
reflect an automatic response to external stimuli. This response is called evoked
potential [9]. In spontaneous interfaces, the user performs a willful cognitive
process or mental task that should be classified [7,10]. The registered EEG signals