Frequency-band specific directed connectivity networks reveal functional disruptions and pathogenic patterns in temporal lobe epilepsy: a MEG study

Abstract

This study investigates the network mechanisms of temporal lobe epilepsy (TLE) using MEG data, focusing on directed connectivity networks across different frequency bands. Unlike previous studies that primarily localize epileptogenic zones, this research aims to explore whole-brain network differences between left TLE (lTLE), right TLE (rTLE), and healthy controls (HCs). MEG data from 13 lTLE patients, 21 rTLE patients, and 14 HCs were source-reconstructed to 116 brain regions (AAL116). Directed Transfer Function (DTF) was used to construct directed connectivity networks, followed by networks and graph-theoretical analyses. The results indicate that, compared to HCs, TLE subjects exhibited a significant increase in average connectivity strength in the Low Gamma band. The connectivity patterns across frequency bands in TLE patients were found to be unstable. Both HC and TLE subjects demonstrated left hemisphere lateralization. In the mid-to-low frequency bands, TLE subjects showed increases in global clustering coefficient (GCC), global characteristic path length (GCPL), and local efficiency (LE) compared to HCs, which is attributed to enhanced synchronization between local brain regions in TLE subjects.

Keywords: Directed transfer function analysis; Graph theoretical analysis; MEG; Temporal lobe epilepsy.

Fig. 1

Workflow for Analyzing Directed Connectivity Brain Networks Across Frequency Bands Using MEG Data.

Fig. 2

Average connectivity strength across different frequency bands for the HCs, lTLE, and rTLE groups. The data represent the mean connectivity values of the normalized DTF-directed brain networks. The x-axis indicates different frequency bands, while the y-axis shows the average connectivity strength values. The box represents the interquartile range (from the first to the third quartile), and the line inside the box indicates the median. The whiskers denote the range of normal values excluding outliers, and each dot represents an individual subject. The SPSS Mann-Whitney U non-parametric test with Holm correction was used to assess the statistical significance of differences between the HCs and the lTLE and rTLE groups, where * indicates , ** indicates , and *** indicates .

Fig. 3

Connectivity patterns across different frequency bands for the three groups of subjects. Each row represents a subject group, and each column represents a specific frequency band. The hub region is highlighted within the red rectangular box. The figure displays only the top 20 connections from the normalized DTF average connectivity matrix. The visualization was generated using the BrainNet Viewer toolbox on the Matlab platform.

Fig. 4

Total outflow of brain regions for HCs, lTLE, and rTLE groups across different frequency bands. This data is obtained by averaging the second dimension of the non-normalized 3D DTF (Directed Transfer Function) connectivity matrix for each group. The x-axis represents different brain regions, while the y-axis represents the corresponding DTF outflow values for those regions.

Fig. 5

DI for HC, lTLE, and rTLE participants across different frequency bands. Figure displays the heatmaps of DI values for each group across different frequency bands. The color bar transitions from blue to red, indicating the directionality of the DI values: red represents a positive DI value (with a maximum of 1), blue represents a negative DI value (with a minimum of -1), and white represents a DI value of 0, indicating no difference.

Fig. 6

DI for HCs, lTLE, and rTLE participants across different frequency bands. Figure shows the difference maps obtained by calculating the differences in DI values between lTLE and rTLE participants compared to HCs across various frequency bands. The color bar transitions from blue to red, indicating the directionality of the DI values: red represents a positive DI value (with a maximum of 1), blue represents a negative DI value (with a minimum of -1), and white represents a DI value of 0, indicating no difference.

Fig. 7

Lateralization Index (LI) for HCs, lTLE, and rTLE groups across different frequency bands. Part a displays the LI values for specific brain regions in each group across various frequency bands. The asterisk (*) indicates the top 10 LI differences obtained by subtracting the LI values of lTLE and rTLE patients from those of HCs for each frequency band, leading to a total of 120 asterisks. Part b shows the overall brain lateralization index for HCs, lTLE, and rTLE patient groups across different frequency bands.

Fig. 8

Four global topological parameters of brain functional networks in HCs, lTLE, and rTLE patients. The SPSS Mann-Whitney U test with Holm correction was used to assess significant differences between HCs and the rTLE and lTLE groups, where * indicates , ** indicates , and *** indicates .

Fig. 9

Workflow for Data Preprocessing.