Eventos Anais de eventos
COBEM 2017
24th ABCM International Congress of Mechanical Engineering
CLASSIFICATION OF RIGHT AND LEFT ELBOWMOVEMENT WITH LOWER GAMMA ACTIVITY
Submission Author:
Elkin Veslin , RJ
Co-Authors:
Max Suell Dutra, Luiz Bevilacqua, Luciano Santos Constantin Raptopoulos, Waltencir Andrade, André da Silva, Mario Fiorani, Juliana Guimarães Martins Soares, Elkin Veslin
Presenter: Elkin Veslin
doi://10.26678/ABCM.COBEM2017.COB17-2456
Abstract
This paper describes the use of lower gamma band for classification of the right versus left elbow rotation in two different cognitive states: planning and movement execution. We made a discrimination analysis for each interval in 14 channels around the motor cortex, and founded common regions of high separability around the cue presentation and after the movement onset and use it to build the feature vector.The dimensionality of the features were reduced with LDA and posteriorly tested with eight classification algorithms in a 10-fold cross validation. The bests results showed that both classes could be identified with a mean of accuracy of 78 % in both intervals with a difference of 2 %. This results shows that the lower gamma band can be used for the classification of elbow movement in BCI systems in specific time intervals using a unique selected band frequency in all subjects.
Keywords
Electroencephalography, Activity recognition, Brain-computer interfaces, Rehabilitation Robotics, machine learning
