Logo

Extracting Single Trial Visual Evoked Potentials using Selective Eigen-Rate Principal Components

Nidal S., Kamel (2005) Extracting Single Trial Visual Evoked Potentials using Selective Eigen-Rate Principal Components. In: 5th International Enformatika Conference, August 2005, Prague, Czech Republic.

[img] PDF
Restricted to Registered users only

153Kb

Official URL: http://citeseerx.ist.psu.edu/viewdoc/download?doi=...

Abstract

Abstract—In single trial analysis, when using Principal Component Analysis (PCA) to extract Visual Evoked Potential (VEP) signals, the selection of principal components (PCs) is an important issue. We propose a new method here that selects only the appropriate PCs. We denote the method as selective eigen-rate (SER). In the method, the VEP is reconstructed based on the rate of the eigen-values of the PCs. When this technique is applied on emulated VEP signals added with background electroencephalogram (EEG), with a focus on extracting the evoked P3 parameter, it is found to be feasible. The improvement in signal to noise ratio (SNR) is superior to two other existing methods of PC selection: Kaiser (KSR) and Residual Power (RP). Though another PC selection method, Spectral Power Ratio (SPR) gives a comparable SNR with high noise factors (i.e. EEGs), SER give more impressive results in such cases. Next, we applied SER method to real VEP signals to analyse the P3 responses for matched and non-matched stimuli. The P3 parameters extracted through our proposed SER method showed higher P3 response for matched stimulus, which confirms to the existing neuroscience knowledge. Single trial PCA using KSR and RP methods failed to indicate any difference for the stimuli.

Item Type:Conference or Workshop Item (Paper)
Subjects:T Technology > TK Electrical engineering. Electronics Nuclear engineering
Departments / MOR / COE:Mission Oriented Research > Health
ID Code:5055
Deposited By: Assoc Prof Dr Nidal Kamel
Deposited On:23 Mar 2011 09:19
Last Modified:19 Jan 2017 08:27

Repository Staff Only: item control page

Document Downloads

More statistics for this item...