A Genetic Algorithm Approach to the Reduction of Additive Noise in Single Trial Evoked Potentials

Authors

  • Michael J. Pougnet University of New Brunswick
  • Dennis F. Lovely University of New Brunswick
  • Philip A. Parker University of New Brunswick

Abstract

A matching tool is introduced for use in noise reduction of single trial evoked potentials. Specific performance markers such as distortion and output SNR are investigated. Results show that the matching tool is able to significantly increase output SNR while reducing the effects of distortion. A new performance metric named the signal improvement quotient is also introduced. This measure represents the ratio of output SNR to distortion. It is suggested that this new metric may be a better measure of noise reduction abilities than a high output SNR. Other factors such as the limitations of the matching tool are also discussed.

Author Biographies

Michael J. Pougnet, University of New Brunswick

BScE, PhD candidate, Department of Electrical & Computer Engineering

Dennis F. Lovely, University of New Brunswick

PhD, PEng, Department of Electrical & Computer Engineering

Philip A. Parker, University of New Brunswick

PhD, PEng, Department of Electrical & Computer Engineering

Downloads

Published

2010-06-15

How to Cite

[1]
M. J. Pougnet, D. F. Lovely, and P. A. Parker, “A Genetic Algorithm Approach to the Reduction of Additive Noise in Single Trial Evoked Potentials”, CMBES Proc., vol. 33, no. 1, Jun. 2010.

Issue

Vol. 33 (2010)

Section

Academic