A comparison of techniques and technologies for SSVEP classification

Author

Tello, Richard M. G. ; Muller, Sandra M. T. ; Bastos-Filho, Teodiano ; Ferreira, Andre

Author_Institution

Post-Grad. Program of Electr. Eng. (PPGEE), UFES, Vitoria, Brazil

fYear

2014

fDate

26-28 May 2014

Firstpage

1

Lastpage

6

Abstract

This paper presents the evaluation of seven techniques of feature extraction (PSD, F-Test, EMD, MCE, CCA, LASSO and MSI) for gaze-target detections in a SSVEP-based BCI. Two type of technologies for visual stimulation were used (LCD and LEDs). Five differents windows lengths (1, 2, 4, 5 and 10 s) were used and seven volunteers participated in this study. The highest accuracy obtained in all cases was 93.57% using LEDs and the highest ITR was 36.90 bits/min for LCD. The technique based on MSI shows the highest success rate in both cases (LCD or LED) and is even more noticeable when the window size is increased.

Keywords

brain-computer interfaces; electroencephalography; feature extraction; medical signal processing; signal classification; synchronisation; visual evoked potentials; CCA; EEG; EMD; F-Test; ITR; LASSO; LCD; LEDs; MCE; MSI; PSD; SSVEP classification; SSVEP-based BCI; brain computer interface; canonical correlation analysis; feature extraction; gaze-target detections; information transfer rate; least absolute shrinkage-and-selection operator; multivariate synchronization index; steady-state visually evoked potentials; visual stimulation; Accuracy; Correlation; Electrodes; Electroencephalography; Feature extraction; Light emitting diodes; Visualization; Brain Computer Interface (BCI); EEG; Steady-State Visual Evoked Potentials (SSVEP);

fLanguage

English

Publisher

ieee

Conference_Titel

Biosignals and Biorobotics Conference (2014): Biosignals and Robotics for Better and Safer Living (BRC), 5th ISSNIP-IEEE

Conference_Location

Salvador

Print_ISBN

978-1-4799-5688-3

Type

conf

DOI

10.1109/BRC.2014.6880956

Filename

6880956