Related Experiment Video
Updated: Jun 25, 2026
09:57
Electromagnetic Source Imaging in Presurgical Evaluation of Children with Drug-Resistant Epilepsy
Published on: September 20, 2024
A neural mass modelling framework for evaluating EEG source localisation of seizure activity
Pok Him Siu1,2, Philippa J Karoly1,2, Sina Mansour Lakouraj1,3
1Department of Biomedical Engineering, The University of Melbourne, Melbourne, Australia.
Journal of Neural Engineering
|June 23, 2026
Summary
Evaluating electroencephalography and magnetoencephalography (EEG/MEG) source localization is challenging without ground truth. A simulation framework with known ground truth shows current methods struggle with polarity, impacting seizure dynamics studies.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Biophysics
Background:
- Objective evaluation of electroencephalography (EEG) and magnetoencephalography (MEG) source localization algorithms is difficult due to the lack of experimental ground truth.
- Characterizing ictal dynamics in epilepsy requires accurate source localization.
Purpose of the Study:
- To present a simulation framework for generating biologically plausible ictal dynamics and corresponding EEG/MEG signals.
- To enable systematic benchmarking of source imaging approaches using simulated data with known ground truth.
Main Methods:
- Simulated cortical seizure initiation and propagation using network-coupled neural mass (Epileptor) models.
- Combined simulated neural activity with realistic forward models of the human head to generate EEG/MEG data.
- Evaluated the MN-family of source localization methods under varying conditions (sensor coverage, noise levels).
Main Results:
- Source localization methods achieved reasonable spatial accuracy in ideal, noise-free conditions.
- Performance degraded significantly with reduced sensor coverage and increased noise.
- Degradation was primarily due to failures in recovering source polarity, even when spatial localization remained accurate.
Conclusions:
- Current EEG/MEG source localization methods may suffice for identifying epileptogenic zones but have limitations in accurately reconstructing polarity.
- Polarity reconstruction is a critical challenge for studying seizure dynamics and network organization.
- The proposed simulation framework offers a reproducible and biologically grounded platform for developing and evaluating source localization techniques.
