getCITED   
  Home     Search     Add Content     Reports     Help  
Edit Publication | Edit Contributors | Delete Publication | Edit References | Edit Citations
Add to Bookstack | Show Bookstack | Change Bookstack

Regressive and Blind Source Separation Techniques for Ocular Artifact Removal

Post a Comment
CONTRIBUTORS:
  Author Theus H. Aspiras
  Author Vijayan K. Asari
JOURNAL:
  International Journal of Information Processing, 6(3), 42 - 54.
YEAR: 2012
PUB TYPE: Journal Article
SUBJECT(S): Electroencephalography, Electrooculography, Eye Artifact Removal, Independent Compo- nent Analysis, Strength of Eye Blink
DISCIPLINE: Computer Science
HTTP: http://www.ijipbangalore.org
LANGUAGE: English
PUB ID: 103-511-665 (Last edited on 2013/01/16 22:18:16 US/Mountain)
SPONSOR(S):
 
ABSTRACT:
Several ocular artifact removal techniques for electroencephalographic data are evaluated in this paper. EEG recordings are taken from an emotion recognition experiment, which contains several instances of ocular artifacts like eye blinks and eye movements. The data is preprocessed through a Butterworth band-pass filter and a 60Hz notch filter to remove most electrical and high frequency noise. Once preprocessed, the data will be used to evaluate three different types of ocular artifact removal techniques: EOG based linear regression, Principal Component Analysis, and Independent Component Analysis. A new metric called Strength of Eye Blink (SEB) is created to automatically determine the removal of different components used in the Blind Source Separation techniques. Each technique is tested using two different metrics: Kurtosis, and a new metric called Zero-Mean Normalized Sum Squared Error. The new metric shows that Independent Component Analysis reduced eye artifacts, the best out of all methods while keeping uncontaminated EEG signals unchanged (Average SSE of 0.1126).
STATISTICS
Click on # to view
 Citations  
 References  
 Comments  
 Quality      0/0.00 
 Interest      0/0.00 
 View(er)s   1/216 
Quality
  N/A
High
  7
  6
  5
  4
  3
  2
  1
Low
Interest
  N/A
High
  7
  6
  5
  4
  3
  2
  1
Low
Prev | Next

    ABOUT getCITED   |    CONTACT US   |    USER INFO   |    PREFERENCES   |    PRIVACY   |    LOG IN   
Comments? Suggestions? Send them to feedback@getCITED.org.

Copyright © 2000-2013 getCITED Inc. All Rights Reserved.