University of Twente Student Theses

Login

Detecting the pain-evoked P300 in single trials: a comparison of template-based models

Portain, D. (2011) Detecting the pain-evoked P300 in single trials: a comparison of template-based models.

[img] PDF
782kB
Abstract:The enhanced Maximum Likelihood method (EML), proposed by Jaskowski and Verleger (2000), is a template-based approach for detecting signals in a continuous EEG measurement. This study compared the ability of three analysis techniques - peak picking, Woody's method and the EML - to determine amplitude and latency of a pain-evoked P3a component. A pseudo-real dataset was constructed from pain-related EEG data and deliberately contaminated with noise. There were two objectives to this research: First, to confirm the validity of our EML implementation; and second, to examine reliability of amplitude estimation in single trials. 15 pseudo-simulated ERP sets were used for the first section. For the second section, amplitude was estimated in four realistic sets of single trials. The original EML was modified twofold: First, a wavelet-transformed signal space was used instead of the Fourier-transformed signal space; and second, the solver used the direct method instead of iterating. Both changes addressed issues (a lack of temporal resolution and oscillating results, respectively) that limited the original study. Comparison of the three techniques yielded results that were consistent with the prior study, validating our implementation. While the EML method proved effective to determine P3a latency in single trials, amplitude estimation was unreliable. The low accuracy of the EML results seems to be caused by the small signal-to-noise ratio. Pain habituation remained undetected in 53% of all subjects, unless single trials were binned in larger chunks.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:77 psychology
Programme:Psychology MSc (66604)
Link to this item:https://purl.utwente.nl/essays/61015
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page