Abstracts

Rationale: Several brain disorders are linked with changes in the resting state network. In previous EEG and fMRI studies, epilepsy patients revealed an increased regularity and order of brain activity compared to normal controls. We investigated whether this holds true for MEG data and assessed the characteristics of epileptic brain networks in idiopathic generalized epilepsy (IGE).Methods: In this preliminary study we measured 8 healthy subjects and 4 IGE patients using whole-head MEG (15 minutes, eyes closed, instructed to stay awake). ECG artifacts were removed using ICA. For patients, interictal discharges (IED) were marked by an experienced clinician. We discarded data segments within a range of 10 seconds before and after the IED to limit our analysis to resting state network in the inter-ictal state. We investigated different frequency bands (delta (1 - 3Hz), theta (4-7Hz), alpha (8-13Hz) and beta (13-28Hz) and broadband activity ranging from 1 to 28Hz). We computed the coherence between each pair of sensors using the imaginary part of coherence to construct a weighted connectivity matrix. Clustering coefficient was calculated for each node. Average shortest path length was calculated per subject based on a graph-theoretical approach (GTA). Subjects networks parameters were averaged across channels. Averages entered a two-sample t-test to test whether the networks in the healthy and epileptic brains were different.Results: Clustering coefficient was significantly higher in delta band (p<0.0007), theta band (p<0.025) and beta band (p<0.029) in IGE patients compared to controls. No significant group difference was found in the alpha band and broadband activity. Shortest path length showed significant increase in all frequency bands including the broadband (delta: p<0.0018, theta: p<0.025, alpha: p<0.0425, beta: p<0.029 and BB: p<0.03). According to graph theory, an ordered graph has a high clustering coefficient and a high shortest path length, while a random graph has a low clustering coefficient and a low shortest path length.Conclusions: Our preliminary results show that patients with IGE have profoundly different networks characteristics compared to controls that indicate more regular and ordered network structure during rest. These results are in accordance with the studies conducted using other modalities like EEG and fMRI. Further analysis is required to study the network in the ictal state, i.e. at the time of spike-wave occurrence compared to the rest conditions, with different epilepsy syndromes and with more subjects. Nevertheless, our results indicate that GTA seems to be a promising method in functional network analysis of epilepsy patients.