A data-driven network analysis strategy was developed to apply EEG-informed functional MRI to identify the seizure onset zone in the presurgical work-up of epilepsy patients (n = 10). Instead of voxel-wise general linear model analysis the time series of independent components were correlated with the interictal epileptic discharges density function, yielding the so-called epileptic network. We used eigenvector centrality mapping and a symmetry index to detect the epileptic independent component (ICE) out of the epileptic network. The location of the ICE was for 9 of the 10 patients studied concordant with the clinical hypothesis. Moreover, the clinical evaluation including the outcome of surgery indicated successful localization of the ICE for 6 out of 8 patients who had a resection. The robustness of the methods used to identify the ICE was demonstrated by evaluating the results of the patient study against the results of similar network analysis procedures applied to the functional MRI sequences of 10 healthy controls. In conclusion, the data-driven network analysis strategy successfully identifies the ICE. The concordance of the ICE with the clinical information, including outcome of the resection of the patients, is in support of the usefulness of EEG-fMRI as initial diagnostic tool in the presurgical work-up of epilepsy patients.
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Conference||1st International Conference on Bioengineering and Biomedical Signal and Image Processing, BIOMESIP 2021|
|Period||19/07/2021 → 21/07/2021|