Graduation Year
2010
Document Type
Dissertation
Degree
Ph.D.
Degree Granting Department
Electrical Engineering
Major Professor
Ravi Sankar, Ph.D.
Committee Member
Emanuel Donchin, Ph.D.
Committee Member
Yael Arbel, Ph.D.
Committee Member
Jing Wang, Ph.D.
Committee Member
Mark Jaroszeski, Ph.D.
Keywords
P300, BSS, ICA, SWLDA, Blind Tracking, Variance Analysis
Abstract
Brain Computer Interface (BCI) is a direct communication channel between brain and computer. It allows the users to control the environment without the need to control muscle activity [1-2]. P300-Speller is a well known and widely used BCI system that was developed by Farwell and Donchin in 1988 [3]. The accuracy level of the P300-BCI Speller as measured by the percent of communicated characters correctly identified by the system depends on the ability to detect the P300 event related potential (ERP) component among the ongoing electroencephalography (EEG) signal. Different techniques have been tested to reduce the number of trials needed to be averaged together to allow the reliable detection of the P300 response. Some of them have achieved high accuracies in multiple-trial P300 response detection. However the accuracy of single trial P300 response detection still needs to be improved. In this research, two single trial P300 response classification methods were designed. One is based on independent component analysis (ICA) with blind tracking and the other is based on variance analysis. The purpose of both methods is to detect a chosen character in real-time in the P300-BCI speller. The experimental results demonstrate that the proposed methods dramatically reduce the signal processing time, improve the data communication rate, and achieve overall accuracy of 79.1% for ICA based method and 84.8% for variance analysis based method in single trial P300 response classification task. Both methods showed better performance than that of the single trial stepwise linear discriminant analysis (SWLDA), which has been considered as the most accurate and practical technique working with P300-BCI Speller.
Scholar Commons Citation
Li, Kun, "Advanced Signal Processing Techniques for Single Trial Electroencephalography Signal Classification for Brain Computer Interface Applications" (2010). USF Tampa Graduate Theses and Dissertations.
https://digitalcommons.usf.edu/etd/3484
