Time domain parameters for online feedback fNIRS-based brain-computer interface systems

Authors:
Tuan Hoang;Dat Tran;Khoa Truong;Trung Le;Xu Huang;Dharmendra Sharma;Toi Vo
Affiliations:
Faculty of Information Sciences and Engineering, University of Canberra, Australia;Faculty of Information Sciences and Engineering, University of Canberra, Australia;Department of Biomedical Engineering, International University, Vietnam;Faculty of Information Sciences and Engineering, University of Canberra, Australia;Department of Biomedical Engineering, International University, Vietnam;Faculty of Information Sciences and Engineering, University of Canberra, Australia;Department of Biomedical Engineering, International University, Vietnam
Venue:
ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part II
Year:
2012

Citing 4
Cited 0

Berlin Brain-Computer Interface-The HCI communication channel for discovery

International Journal of Human-Computer Studies
Person Authentication Using Brainwaves (EEG) and Maximum A Posteriori Model Adaptation

IEEE Transactions on Pattern Analysis and Machine Intelligence
Unobtrusive biometric system based on electroencephalogram analysis

EURASIP Journal on Advances in Signal Processing
Identification of moving limb using near infrared spectroscopic signals for brain activation

IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

We investigate time domain parameters called high order moments in functional Near Infrared Spectroscopy (fNIRS) signal and propose to use them as new brain features in fNIRS-based Brain Computer Interface (BCI) research. These high order moments are well appropriate with fNIRS data without any special preprocessing or filtering step. Therefore, they could be used to guide users in feedback fNIRS-based BCI experiments. We performed experiments on motor imagery and person identification problems with the 2nd order moment, 4th order moment and a combination of these moments. Experimental results showed that these features provided high accuracy. For motor imagery problem, our system could achieve accuracy up to 99.5% for subject independent problem and varies between 86.5α5.4% and 97.0α2.1% for subject dependent problem. For person identification problem, our system could achieve accuracy nearly 100%. Comparing with other systems that used non-filtered raw signal as feature, these features are more stable than the raw signal because of noise reduction. We also found that the 2nd order moment alone could be an excellent and efficient feature for fNIRS-based BCI systems.