Skip to content

esmondo/masterthesis_decoding

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 

Repository files navigation

MATLAB Scripts for a Master Thesis:

Single Trial Decoding of Visual Spatial Attention Shifts from EEG and MEG Signals

Master's thesis written by Ardiansyah Esmondo to obtain the academic degree of Master of Science (M.Sc.) from Otto-von-Guericke University in the master's program of Medical Systems Engineering.

Abstract

Purpose | Humans are constantly shifting their attention from one item to another. It has always been found that during a visual search task where attentional processing discriminates target and distractors, a particular event-related potential (ERP) component is elicited, known as N2pc. This thesis investigates whether visual-spatial attention shifts can only be extracted from a grand ERP average or decoded from a single trial.

Method | To investigate the decodability at the single-trial level, binary classification of the EEG and MEG signals associated with the spatial attention shifts was employed by predicting the position of the attended target. The decoding accuracy is the number of correctly predicted target classes out of all the predictions. Implementing a spatiotemporal filter (STF) and generating combined trial datasets were included in the classification to increase the signal-to-noise ratio (SNR).

Results | The obtained decoding accuracy reflected that the EEG and MEG associated with the N2pc were reliable to be decoded at the single-trial level. Mean decoding accuracy was reported to be higher with STF implementation and constantly increased as more combined-trial datasets were added. At the single-trial level, the mean decoding accuracy of EEG and MEG was 67.9% and 68.8%, respectively, and with STF implementation was 73.2% and 69.5%, respectively.

Conclusion | Stronger discriminability to a relevant item among distractors in both visual fields could lead to more reliable N2pc in a single trial EEG/MEG, which drives better decoding performance.

Keywords | Event-related Potentials, Attention shifts, N2pc, Selective Attention, Decoding, Classification, Support Vector Machine, Electroencephalography, Magnetoencephalography.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors