Paul Pawletta, BCCN Berlin

Realistic and Template Head Models for EEG Source Localization

State of the art head models for EEG source localization incorporate knowledge about subject specifc head anatomy, for example from MR images. However, obtaining MR images in advance is time and cost consuming and is often omitted. But utilizing template head models instead of the realistic head model introduces a source localization error. This work compared different MNI template head models for source localization and aims to enhance the accuracy of template head models. A method to correct source localization for template head models was derived from statistics of simulated dipole sources of a small subject population. It could be shown that the correction applied to simulated as well as real EEG sources yielded source location estimates which were closer to the estimates of the realistic head model. In addition, confidence ellipses were estimated for template head models to highlight brain regions that are likely to contain the source location estimates of the realistic head model. 

Additional Information

Master thesis defense as part of the MSc Computational Neuroscience

Organized by

Benjamin Blankertz / Robert Martin