Classification of real and imaginary hand movements for a BCI design

Author

Ozmen, Nurhan Gursel ; Gumusel, Levent

Author_Institution

Mech. Eng. Dept., Karadeniz Tech. Univ., Trabzon, Turkey

fYear

2013

fDate

2-4 July 2013

Firstpage

607

Lastpage

611

Abstract

This paper searches the discrimination ability of a feature extraction method for EEG analysis. The method is tested on the classification of imagined and real right/left hand movements. The study serves for brain computer interface (BCI) applications which help people to control their body via thoughts. According to the results of the three different classifiers which are LDA, SVM and NN, it is concluded that imagination of hand movements can be used instead of real hand movements especially for tetraplegic patients. The classification accuracies of imaginary hand movements of two subjects are 96% and 99% and accuracies of real hand movements are 85% and 77% respectively.

Keywords

electroencephalography; feature extraction; medical signal processing; neural nets; signal classification; support vector machines; BCI design; EEG analysis; LDA; NN; SVM; brain computer interface; feature extraction method; imaginary hand movement classification; imagined right-left hand movements; linear discriminant analysis; neural networks; real hand movement classification; real right-left hand movements; support vector machines; tetraplegic patients; Accuracy; Artificial neural networks; Brain-computer interfaces; Electrodes; Electroencephalography; Feature extraction; Support vector machines; EEG; Feature extraction; LDA; NN; SVM; motor task classification;

fLanguage

English

Publisher

ieee

Conference_Titel

Telecommunications and Signal Processing (TSP), 2013 36th International Conference on

Conference_Location

Rome

Print_ISBN

978-1-4799-0402-0

Type

conf

DOI

10.1109/TSP.2013.6614007

Filename

6614007