Breakdown of Thalamo-Cortical Connectivity Precedes Spike Generation in Focal Epilepsies

Abstract

Electroencephalography (EEG) spikes and focal epileptic seizures are generated in circumscribed cerebral networks that have been insufficiently described. For precise time and spatial domain network characterization, we applied in patients with focal epilepsy dense array 256-channel EEG recordings with causal connectivity estimation by using time-resolved partial directed coherence and 3T-magnetic resonance imaging-derived cortical and thalamus integrity reconstruction. Before spike generation, significant theta and alpha bands driven information flows alterations were noted from both temporal and frontal lobes to the thalamus and from the thalamus to the frontal lobe. Medial dorsal and ventral anterior nuclei of the thalamus were delimited as possible pacemakers. Markedly reduced thalamic volumes and impaired cortical integrity in widespread areas predicted the altered information flows. Our data reveal distinct patterns of connectivity involving the thalamus and frontal cortex that are both directly and causally involved in spike generation. These structures might play an essential role in epileptogenesis and could be targeted in future therapeutic approaches.

Introduction

Focal epileptogenesis relies on perturbations of neuronal activity regionally and in a network of interconnected anatomical regions. Modified excitatory/inhibitory interactions within cortico-cortical or subcortical networks potentiate the local activity, generating long-range synchronizations. Pathologic synchronization could lead to interictal spikes, which are bursts of action potentials that are generated by recurrent excitation, followed by a period of inhibition (McCormick and Contreras, 2001). This synchronization of electric activities recorded noninvasively is, therefore, an essential phenomenon that could help to resolve epileptogenic networks (Bartolomei et al., 2001; de Curtis and Avanzini, 2001). Recent studies investigating pathological synchronous neuronal activity and its structural correlates have led to a paradigm shift from “epileptogenic focus” to “epileptogenic network” (Richardson, 2012; Siniatchkin et al., 2007; Yaffe et al., 2015). One of the remote structures of the epileptogenic networks that is involved early in the temporal and spatial synchronization leading to seizure generation might be the thalamus; however, its precise role in the propagation of focal seizures is not clear (Groppa et al., 2008, 2012, 2013; Rosenberg et al., 2006). The thalamus as a network node not only receives substantial neuronal input from different brain regions but has also strong reciprocal connections to other subcortical and cortical areas (Groppa et al., 2008, 2012). Insausti et al. (1987) showed that the mediodorsal nucleus of the thalamus has rich reciprocal interconnections with neocortical areas, such as with the frontal and prefrontal cortex (PFC) through the anterior thalamic peduncle (Tanaka, 1976) and with temporal regions such as the amygdala and olfactory cortex (Russchen et al., 1987) through the inferior thalamic peduncle. These pathways might be of special importance for seizure generation. As recently shown (Bernhardt et al., 2012; Guye et al., 2006; Mueller et al., 2010), the thalamus might also be involved in the long-range coupling in temporal lobe epilepsy (TLE), the most common type of focal epilepsy. However, it is still not known as to which thalamic subregions in humans modulate the information flows related to epileptogenesis and which frequency bands are involved. For the development of new therapeutic techniques such as invasive or noninvasive stimulation or neurosurgical methods, it is of utmost importance to characterize these functional and structural abnormalities and describe the entire network involved (Groppa et al., 2014; Muthuraman et al., 2016). In the adopted approach of structural and functional network characterization, we aimed at obtaining an exact and causal estimation of the temporal and spatial properties of the involved networks. Moreover, the adopted approach describes the directionality of the information flows and achieves an exact temporal dissection of the connectivity dynamics. The proposed frequency domain analysis has a unique feature to describe time- and space-specific processes linked to distinct frequency bands that drive the long-range connectivity.

We hypothesize (1) frequency- and time domain-specific changes in connectivity before spike generation and (2) clear structural abnormalities involving the epileptogenic network that are related to the long-range synchronization and not solely to the primary focus.

Methods

Ethics statement

The study protocol was approved by the local ethical board at the National Center of Epileptology institutional review board, Chisinau, Moldova.

Participants

Fifteen patients (mean age ± standard deviation: 28.1 ± 8.3 years, 9 men) with focal epilepsy (12 TLE patients) without seizures in the month before investigation and 15 age-matched healthy subjects (27.9 ± 4.0 years, 7 men) were included. The diagnosis of epilepsy was established on the basis of clinical history and electroencephalography (EEG) findings according to revised International League Against Epilepsy (ILAE) classification (Berg et al., 2010). The recruited patients were referred to the National Center of Epileptology, Chisinau, Moldova. The demographic and clinical characteristics of each patient are summarized in Table 1. The analysis pipeline is presented in Figure 1.

FIG. 1.

Overview of the analysis pipeline: (a) electroencephalography (EEG) processing stream; (b) magnetic resonance imaging (MRI) processing stream. TPDC, time-resolved partial directed coherence. Color images available online at www.liebertpub.com/brain

Table 1.

Demographic, Clinical, and Magnetic Resonance Imaging Data of Epilepsy Patients

Patient No.	Handedness	Age	Gender	Age at onset	Seizure type	MRI findings	AEDs
1	R	22	M	15	DS→BCS	Normal	CBZ + LCM
2	R	24	F	17	DS→BCS	R temporoparietal GMH	LTG
3	R	14	M	6	DS→BCS	L MTS	CBZ
4	R	32	F	10	EMS→BCS	L MTS	CBZ + LTG
5	L	15	M	2	EMS→BCS	L parieto-occipital EM	LTG
6	R	27	F	9	DS→BCS	R temporal glial scar	CBZ
7	R	26	M	12	DS→BCS	L temporal glial scars	CBZ + LTG
8	R	32	M	15	DS→BCS	Nonspecific WM lesions	CBZ + LTG
9	R	32	M	11	DS→BCS	L MTS	CBZ + LEV
10	R	22	F	13	DS→BCS	L MTS	CBZ + LEV
11	R	31	F	7	DS→BCS	R temporal subarachnoid cyst	CBZ + LTG
12	R	26	F	9	EMS→BCS	R MTS	CBZ
13	R	38	M	35	DS→BCS	Nonspecific WM lesions	CBZ
14	R	36	F	2	DS→BCS	L MTS	CBZ + LTG
15	R	45	M	41	DS→BCS	WM, L PHG foci of demyelination	VPA

AEDs, antiepileptic drugs; BCS, bilateral convulsive seizures; CBZ, carbamazepine; DS, dyscognitive seizures; EM, encephalomalacia; EMS, elementary motor seizures; GMH, gray matter heterotopia; L, left; LCM, Lacosamide; LEV, levetiracetam; LTG, lamotrigine; MRI, magnetic resonance imaging; MTS, mesial temporal sclerosis; PHG, parahippocampal gyrus; R, right; WM, white matter, VPA, valproate.

EEG data acquisition

We performed interictal dense array video EEG of epilepsy patients at rest and in an alert state for 2 h in a dimly lit and quite room. The EEG electrodes were included in a cap, with 20–25 mm interelectrode distances placed according to the international 10/5 system (Jurcak et al., 2007) (Hydrogel Geodesic Sensor Net 130 routine, 256 electrodes, Electrical Geodesic, Inc., Eugene, OR). The sampling rate was 1000 Hz, with low and high frequency filters (0.3 and 70 Hz, respectively) and stored using Net Station 5 software package (Electrical Geodesic) and in reference to anatomical landmarks. The sampling rate was 1000 Hz. Net station 5 software package (Electrical Geodesic) was used to filter the data at 0.3 Hz high pass and 70 Hz low pass filters respectively. The obtained electrodes’ impedance was below 10 kΩ. All patients were under continuous observation throughout the recording.

Magnetic resonance imaging data acquisition

Structural magnetic resonance imaging (MRI) data were acquired by using a 3T SIEMENS Skyra scanner, with a 32-channel head coil (Wellmer et al., 2013). T1-weighted images (repetition time [TR] = 2000 msec, echo time [TE] = 9 msec, 4 mm slice thickness, flip angle = 150°), T2-weighted images (TR = 3800 msec, TE = 117 msec, 4 mm slice thickness, flip angle = 190°), and fluid-attenuated inversion recovery sequences were acquired.

Data processing

EEG preprocessing and analysis was performed by using FieldTrip toolbox and in-house Matlab scripts. Muscle artifacts were first removed by using manual data inspection by an experienced neurophysiologist. An infomax Independent Component Analysis (ICA) (Delorme and Makeig, 2004) was then applied and ICA components were profiled by their topography, activation time course, and spectrogram. Components clearly assigned to, for example, rarely occurring eye blinks (Jung et al., 2000) were excluded from the back projection. EEG spikes from all scalp channels were then transformed to the average reference (Lehmann and Skrandies, 1980), and EEG segments with remaining artifacts were removed. Typical interictal spikes with the same morphology were marked manually at the time of maximum positivity or negativity, the detected spikes were checked individually, and some were excluded. In turn, interictal events such as polyspikes or spikes riding on a preceding slow wave were also rejected to avoid preonset baseline contamination. For each patient, 5–10 interictal spikes were selected. The EEG signal was parsed 15 sec before the spike peak for each patient separately. The time interval of 10 sec before the spike was defined as the activation period, and the time period from 10 to 15 sec before the spike was defined as the baseline period for all the following analyses. Source power and synchronization analysis (see section Source analysis) across all sources was examined at theta (4–7 Hz), alpha (8–13 Hz), and beta (14–30 Hz) frequency bands separately.

Source analysis

A full description of the coherence analysis is given elsewhere (Michels et al., 2013; Muthuraman et al., 2012). There are two major constraints in this analysis: First, the analysis is created on a single dipole model, which is not linearly correlated to other dipoles and second, the signal-to-noise ratio must be sufficiently high (Groß et al., 2001). The fixed dipole model was used in which the dipole source responsible for the measured EEG potentials during an epoch remains at a constant location, the dipole moment vector maintains a constant orientation throughout the epoch, and only the magnitude varies. The output of the beamformer at a voxel in the brain can be defined as a weighted sum of the output of all EEG channels (Van Veen et al., 1997). The weights determine the spatial filtering characteristics of the beamformer and are selected to increase the sensitivity to signals from a voxel and reduce the contributions of signals from (noise) sources at different locations. The frequency components and their linear interaction are represented as a cross-spectral density matrix. To visualize power at a given frequency range, a linear transformation was used based on a constrained optimization problem, which acts as a spatial filter (Van Veen et al., 1997). The spatial filter was applied to a large number of voxels covering the entire brain, assigning a specific value of power to each voxel. A voxel size of 5 mm was used in this study. The beamformer weights for a given source (at a location of interest) are determined by the data covariance matrix and the forward solution (lead-field matrix [LFM]). The LFM was estimated with specified models for the brain. In this study, the brain was modeled by a boundary element method (Fuchs et al., 2002) with three layers, namely brain, skull, and skin. The volume conductor model was created by using individual T1 magnetic resonance images. Part of the forward modeling and the source analysis were done by using the open source software FieldTrip (Oostenveld et al., 2011). For both groups, the head was modeled by using the electrode locations based on the anatomical landmarks. The LFM contains information about the geometry and conductivity of the model. The complete description of the solution for the forward problem has been previously described (Muthuraman et al., 2010, 2014).

The brain regions were defined according to the Automatic Anatomic Labeling Atlas 116 regions of interest (ROIs) (Whitfield-Gabrieli and Nieto-Castanon, 2012). To address major anatomically relevant regions, the ROIs were subdivided into seven separate regions, namely prefrontal, frontal, parietal, occipital, temporal, thalamus, and cerebellum. The thalamus parcellation was done based on the connectivity probability to cortical masks (Behrens et al., 2003; Jones, 2012). The activated voxels were selected by a within-subject surrogate analysis to define the significance level, which was then used to identify voxels in the regions as activated voxels. Once brain region voxels were identified, their activity was extracted from the surface EEG (source space). In a further analysis, all the original source signals for each region with several activated voxels were combined by estimating the second-order spectra and employing a weighting scheme depending on the analyzed frequency range to form a pooled source signal estimate for each region as previously described (Amjad et al., 1997; Rosenberg et al., 1989). For estimating the relative signal-to-noise ratio (SNR), the power in the frequency bands (theta, alpha and beta) was considered as signal and the power in the frequency band of 100–200 Hz from the averaged epochs (5–10 epochs per patient) was considered as noise. We have also estimated the orientation of dipoles in the thalamus voxels to justify the dipole orientation. From the number of activated voxels the principle orientation was calculated and normalized by the orientations from the total number of activated voxels.

Connectivity analysis

Using time-frequency causality, we can not only focus on a particular frequency but also analyze the time dynamics of the causality at that frequency. Based on the state-space modeling, the time-frequency causality estimation method of time-resolved partial directed coherence (TPDC) utilizes dual-extended Kalman filtering (DEKF) (Haykin, 2001; Wan and Nelson, 2001). We can estimate the time-varying dependent autoregressive (AR) coefficients. One extended Kalman Filter (EKF) will estimate the states and feed this information to the other EKF; the second EKF will estimate the model parameters and will also share this information with the first EKF. By concurrently using two Kalman filters, working in parallel with one another, we can estimate both states and model parameters of the system at each time instance. After estimating the time-varying multivariate autoregressive (MVAR) coefficients, the next step is to use those coefficients for the calculation of causality between the time series. Since DEKF can give us the time-varying MVAR coefficients at each time point, we can calculate the partial directed coherence (PDC) at each time point as well. The general expression for the PDC is given as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} \left\vert { { \pi _ { i \leftarrow j } } ( \omega ) } \right\vert = { \frac { \left\vert { { A_ { ij } } ( \omega ) } \right\vert } { \sqrt { \sum \nolimits_k { { { \left\vert { { A_ { kj } } ( \omega ) } \right\vert } ^2 } } } } } \tag { 1 } \end{align*} \end{document}

In Equation (1), π is the magnitude of PDC from time series j to i at frequency \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\omega$$ \end{document} , which are the Fourier transform coefficients of the causal coefficients. Afterward, time-frequency plots of all PDCs can be concatenated to produce a single time-frequency plot. The precise distribution of the MVAR coefficients is not known; we used the surrogate method to check for the significance of the results. This method is based on the random shuffling of the subjected time series and, hence, it is data driven. In short, we divided the original time series into smaller nonoverlapping windows and randomly shuffled the order of these windows to create a new time series. The MVAR model is fitted to this shuffled time series, and the TPDC is estimated. This process is repeated 100 times, and the average TPDC is calculated. The resulting value is the significance threshold value for all our connections. This process is performed separately for each patient.

The mean TPDC values for the significant connections from the different time windows were compared with a repeated-measure analysis of variance. The significance level was set to p < 0.05. Post hoc tests with Tukey-Kramer correction for multiple comparisons were applied.

Cortical thickness analysis

Cortical reconstruction and volumetric segmentation were performed with the Freesurfer image analysis suite (version 5.3.0), which is documented and freely available for download online (http://surfer.nmr.mgh.harvard.edu/). The detailed procedure for surface and volumetric reconstruction has been described and validated in previous studies (Dale et al., 1999; Fischl and Dale, 2000; Reuter et al., 2012). In brief, cortical thickness was calculated (in mm) as the shortest distance between gray matter/white matter boundary and pial surface across the cortical mantle. Thalamus volume was calculated by using the automated procedure and checked visually. Cortical thickness was smoothed with a 10-mm full-width at-half-height Gaussian kernel to reduce local variations in the measurements (Du et al., 2007). Statistical difference was computed by using a random-effects model with t-tests for each cortical location. For statistical difference maps, the significant threshold was set to an uncorrected p-value of p < 0.05 (two tailed) (Lyoo et al., 2006).

To relate information flow changes and structure, we computed the difference ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\nabla = {\rm max}_{Tpdc} - {\rm min}_{Tpdc}$$ \end{document} ) in connectivity dynamics before spike generation and cortical thickness and thalamic volume values.

Results

Clinical description

In this study, a total of 15 epilepsy patients and 15 age-matched healthy controls were analyzed. The healthy controls did not present any history of seizures or other neurologic abnormalities. There was no significant difference in age (p > 0.05) between the two cohorts. All patients except one were right handed. Thirteen patients had dyscognitive seizures, and three patients presented focal motor seizures with evolution in bilateral convulsive seizures. Among 15 patients, the radiological magnetic resonance imaging (MRI) assessment showed no morphological correlate in one case, and the others presented different structural lesions (Table 1).

Temporal dynamics of effective connectivity

The analysis of the focused frequency bands (theta, alpha, and beta) in the period before spike generation showed significant information flows only in theta (t = 4.06, p = 0.005) and alpha (t = 4.36, p = 0.003) frequency bands. These survived the data-driven surrogate significance analysis and will be discussed further in this section.

In patients, we detect a significant decrease in connectivity strength from temporal and frontal lobes to the thalamus and from the thalamus to the frontal lobe in the theta domain with a maximum peak 5 sec before spike generation (Fig. 2a). Moreover, just before spike appearance (in the last 4 sec), the connectivity from the temporal lobe to the thalamus increased significantly. Furthermore, we observed a similar pattern of connectivity dynamics in the alpha band (Fig. 2b). Here, we noticed a clear information flow increase from the temporal lobe to the thalamus that also peaked 5 sec before spike generation. A further significant breakdown of information flows between the temporal and frontal lobe to the thalamus was noted.

FIG. 2.

Temporal dynamics of effective connectivity in theta (a) and alpha (b) frequency bands. Connectivity breakdown between 4 and 5 sec before the spike in reciprocal interconnections from the frontal lobe to the thalamus and from the thalamus to the frontal lobe. Significant changes of information flows at *p < 0.001, **p < 0.0001. Color images available online at www.liebertpub.com/brain

The strength of temporo-thalamic connectivity change (5 sec before spike) significantly correlated with fronto-thalamic connectivity dynamics (r = 0.73, p = 0.001). In the same band and time window, the thalamo-frontal connectivity values showed no significant interrelations with temporo-thalamic (p = 0.06) and fronto-thalamic (p = 0.5) connectivity variations.

In the alpha band, thalamo-frontal connectivity correlated significantly with the modulation of fronto-thalamic (r = 0.63, p = 0.01) and not significantly with the temporo-thalamic (p = 0.7) information flows. Temporo-thalamic connectivity also did not correlate with the fronto-thalamic connectivity dynamics (p = 0.5).

No other significant changes of connectivity were noted between the further analyzed cerebral regions (i.e., occipital, parietal, prefrontal).

In healthy subjects for the analyzed frequency bands (theta, alpha, and beta), we found significant information flows only in the alpha frequency band. Bidirectional connections were attested between the frontal lobe and the thalamus (t = 4.01, p = 0.006) and between frontal and parietal lobes (t = 4.16, p = 0.004) that survived the data-driven surrogate significance analysis. In the analyzed time window of 10 sec, no significant changes in information flows were detected for these connections, which were stable over the entire time interval.

Thalamic parcellation-related connectivity dynamics

In the theta frequency band, we found significant bidirectional connections from the medial dorsal and ventral anterior thalamus to the frontal lobe (Fig. 3). The temporal dynamics of connectivity changes were similar to the changes of connectivity involving the entire thalamus structure mentioned earlier. In the alpha frequency band, we see a similar breakdown of the connectivity strength involving the frontal lobe and the medial dorsal and ventral anterior nuclei of the thalamus (Fig. 4). A significant reduction in information flows was detected ca. 5 sec before spike generation from the temporal lobe to the anterior, lateral, and inferior nuclei of the pulvinar of the thalamus, as shown in Figures 5 and 6. The relative SNR for each of the frequency bands was above 30 decibel (dB), and only those epochs were chosen which had a relative SNR greater than 30 dB and the orientation of dipoles was in all the subjects radial (above 75%) that additionally supports the argument for visualizing deeper thalamic sources (see Supplementary Table S1; Supplementary Data are available online at www.liebertpub.com/brain).

FIG. 3.

Connectivity dynamics in theta frequency band from the frontal lobe to the medial dorsal nucleus of the thalamus (a) and vice versa (b) and from the frontal lobe to the ventral anterior nucleus of the thalamus (d) and vice versa (e). Connectivity breakdown between 4 and 5 sec before the spike (*p < 0.001, **p < 0.0001). Reciprocal flows between the frontal lobe and the medial dorsal nucleus (brown colored structures) (c) and between the frontal lobe and the ventral anterior nucleus (blue-colored structures) (f) of the thalamus. Color images available online at www.liebertpub.com/brain

FIG. 4.

Connectivity dynamics in alpha frequency band from the frontal lobe to the medial dorsal nucleus of the thalamus (a) and vice versa (b) and from the frontal lobe to the ventral anterior nucleus of the thalamus (d) and vice versa (e). Connectivity breakdown between 4 and 5 sec before the spike (*p < 0.001, **p < 0.0001). Reciprocal flows between the frontal lobe and the medial dorsal nucleus (brown-colored structures) (c) and between the frontal lobe and the ventral anterior nucleus (blue-colored structures) (f) of the thalamus. Color images available online at www.liebertpub.com/brain

FIG. 5.

Effective connectivity dynamics in theta frequency band from the temporal lobe toward anterior (a), lateral (b), and inferior (c) nuclei of the pulvinar of the thalamus. Directional flows from the temporal lobe toward lateral and inferior (green-colored structures) (d) and anterior (orange-colored structures) (e) pulvinar nuclei of the thalamus. *p < 0.0001. Color images available online at www.liebertpub.com/brain

FIG. 6.

Effective connectivity dynamics in alpha frequency band from the temporal lobe toward anterior (a), lateral (b), and inferior (c) nuclei of the pulvinar of the thalamus. Directional flows from the temporal lobe toward lateral and inferior (green-colored structures) (d) and anterior (orange-colored structures) (e) pulvinar nuclei of the thalamus. *p < 0.0001. Color images available online at www.liebertpub.com/brain

Cortical thickness and thalamic volume

Cortical thickness analysis showed a significant cortical thinning of widespread regions in patients with epilepsy (p < 0.05, false discovery rate corrected; >50 voxels), mainly in the frontal and temporal regions (Fig. 7). A cortical thickness increase was detected in the occipital cortex. The complete results are presented in Table 2 and Figure 7. No significant correlations were found between cortical thickness and disease duration.

FIG. 7.

Statistical maps showing clusters (p < 0.05, false discovery rate corrected) of cortical thinning (blue) and thickening (red) in epilepsy patients compared with healthy subjects. The two slices in (a) show the regions in the right hemisphere and (b) shows those in the left hemisphere. Color images available online at www.liebertpub.com/brain

Table 2.

Significant Clusters Derived from Comparison Between Focal Epilepsy Patients and Healthy Controls, Tabulated with the Corresponding t-Values, Surface Area (mm ²), and Talairach Coordinates

Right hemisphere
Cluster No.	t-Max	Size (mm²)	Tal X	Tal Y	Tal Z	Annotation
1	−6.7544	1475.37	60.7	−41.4	17.4	Supramarginal
2	−6.4630	412.76	58.1	−49.1	3.3	Bankssts
3	−5.0150	820.58	62.1	−11.2	−22.2	Middletemporal
4	−4.6995	307.02	52.7	6.6	−13.5	Superiortemporal
5	−4.3071	272.79	29.7	6.1	−33.2	Temporalpole
6	−4.2364	439.18	36.8	−0.9	−7.0	Insula
7	3.9993	317.45	5.0	−84.6	9.5	Cuneus
8	−3.9980	372.07	7.0	36.2	46.9	Superiorfrontal
9	−3.8235	103.50	58.2	0.6	28.8	Precentral
10	−3.6384	108.07	5.9	−36.0	2.0	Posteriorcingulate
11	−3.4420	182.46	2.6	12.3	4.6	Parsopercularis
12	3.2205	167.34	13.5	37.0	−19.5	Lateralorbitofrontal
13	3.2088	319.11	6.5	−76.7	−0.8	Lingual
14	−3.1768	49.43	11.2	−63.7	61.2	Superiorparietal
15	−3.1348	63.97	38.3	−33.4	53.9	Postcentral
16	−2.9896	25.80	47.5	33.5	−12.	Parsorbitalis
17	−2.8765	100.05	44.4	−69.0	6.1	Inferiorparietal
18	2.8507	60.53	19.1	−97.1	−13.3	Lateraloccipital
19	−2.7798	88.59	6.9	−23.0	55.4	Paracentral
20	−2.7326	26.22	6.6	−50.4	19.8	Isthmuscingulate

Left hemisphere
1	−5.4614	192.61	−57.3	−0.5	28.8	Precentral
2	−6.7544	273.72	−57.0	44.1	32.6	Supramarginal
3	−3.7923	18.66	−52.5	−19.3	52.6	Postcentral
4	−3.7650	50.19	−39.4	4.2	51.6	Caudalmiddlefrontal
5	−3.5873	20.10	−6.4	42.7	45.4	Superiorfrontal

Thalamic volumes were markedly reduced in epilepsy patients compared with healthy subjects. Volumes of left (7362.1 ± 848.3 mm³, p = 0.00005) and right (7186 ± 848.3 mm³, p = 0.0037) thalami of FE patients were significantly smaller in comparison with healthy controls (left thalamus 9360.5 ± 1382 mm³, right thalamus 8088.7 ± 683 mm³).

Correlation effective connectivity and thalamic volume

In the theta band, the temporo-thalamic effective connectivity significantly correlated with the volumes of both thalami (r = 0.8981, p < 0.001), whereas the fronto-thalamic connectivity did not show significant correlations with the thalamic volumes (p > 0.1) or only a trend for the thalamo-frontal connection (r = −0.5, p = 0.058). In the alpha band, only the fronto-thalamic effective connectivity significantly correlated with the volumes of both thalami (r = −0.61, p < 0.01); whereas the correlations between temporo-thalamic and thalamo-frontal effective connectivities and thalamic volumes did not reach statistical significance (p > 0.1 and p > 0.5 respectively).

Discussion

In this work, we analyze information flows and structural integrity as network fingerprints of the circuits that drive the spike generation and seizure evolvement. Establishing the directionality and the causality of cerebral dynamics is essential for linking network connectivity and seizure prediction. Therefore, we investigated noninvasively the alterations of effective connectivity in a group of patients with focal epilepsy during the interictal period and specifically focused on the narrow period before spike appearance on scalp EEG. Moreover, we dissected the cortical and subcortical integrity within these networks that modulate the long-range information flows and synchronizations. We see clear connectivity alterations in the theta and alpha bands before spike generation. Our analysis addressed bidirectional information flows among cortical regions and thalamus. Significant information flows were detected in a network formed by temporal and frontal cortices and thalamus. The performed analyses demonstrate a clear breakdown of connectivity patterns 5 sec before spike generation, with a subsequent increase. Medial dorsal and ventral anterior nuclei of the thalamus were mainly involved. Widespread cortical integrity abnormalities and thalamic atrophy were attested in the patient group.

Frequency band specificity

The main effective connectivity variations were identified in theta and alpha bands. A clear physiological framework explains our results. Theta oscillations synchronize the mesial temporal lobe (MTL) with distant cortical and subcortical regions (Bettus et al., 2008; Zaveri et al., 2001), whereas pathological synchronization phenomena involving MTL could be as well mirrored in this frequency band. The underlying cause of these connectivity alterations has not yet been described. One putative explanation could be that the breakdown of effective connectivity from temporal and frontal lobes toward the thalamus in the theta band is caused by an altered physiological synchronization. The state of decreased synchronization could lead to vulnerability to following information flows, increasing the probability of spikes or seizures (Mormann et al., 2003). Consequently, we see an increase of information flows in the temporo-thalamic connections thereafter. We did not account for the influence of antiepileptic drugs on attested connectivity alterations. Lamotrigine and valproate could modify the cross-talks between regions and effective connectivity values, as shown between primary motor and pre-motor cortex (Li et al., 2011). Further studies could consider EC alteration in dependency of blood levels of antiepileptic drugs (Li et al., 2011).

A recent study investigating information flows in patients with TLE by the aid of EEG-functional MRI analysis (Faizo et al., 2014) showed similarly significant reductions in information flows before spike occurrence. A decreased connectivity between both temporal lobes has been attested before spike generation. Summated, it can be postulated that putative breakdowns of information flows in the involved structures lead to vulnerable states of decreased synchronization that can drive epileptogenesis.

Region specificity

In our study, we see a direct involvement of the thalamus, temporal and frontal lobes, sparing occipital and parietal cortices and other studied areas. The described information flows involving the thalamus utilize physiological circuits. Different thalamic nuclei are involved and modulate long-range synchronizations at distinct frequency bands (Wang, 2010). The anterior thalamus is involved in theta synchronization, whereas the pulvinar is responsible for alpha-driven information flows (Ketz et al., 2015). Through theta synchronization, the anterior thalamus modulates the information flows between the hippocampus and medial PFC, whereas pulvinar nuclei synchronize the parietal and visual cortex with medial, lateral, and orbital PFC within the alpha band (Ketz et al., 2015). In our analysis related to thalamic nuclei and temporo-thalamo-frontal connections, the anterior nucleus of the thalamus displayed significant alterations of effective connectivity in theta and alpha ranges, whereas the medial dorsal nucleus showed long-range synchronizations in theta and alpha bands, but not in beta. Lateral and inferior nuclei of the pulvinar showed a connectivity increase in theta and a decrease in alpha synchronization range.

The described connectivity dynamics show some similarities to known network determinants in generalized epilepsies. In this type of epilepsies, fronto-thalamic synchronizations are typical hallmarks of very early absence of seizure activity, as evidenced by EEG and MEG (Groppa et al., 2008, 2012; Holmes et al., 2004; Tenney et al., 2013). Disruptions of synchrony-limiting mechanisms in the thalamus might lead to hypersynchrony with cortical areas and predispose to spike-wave discharges and absence of seizures (Paz et al., 2011).

Some of the regions identified in our study such as medial temporal and frontal lobes are a part of the so-called default mode network (DMN). Studies analyzing structural and functional connectivity abnormalities within the medial temporal lobe and DMN brain areas with (Kay et al., 2013; Pittau et al., 2012) or without (Morgan et al., 2012; Voets et al., 2012) simultaneous EEG recording suggest the presence of important disconnections in DMN sites in TLE patients. Moreover, interictal discharges might be associated with significant deactivation in default mode brain areas (Laufs et al., 2007).

In the comparison of gray matter structural integrity between patients and healthy controls, we depict widespread areas in not only the temporal and frontal regions but also parietal and PFC, suggesting a putative relationship with long-range synchronization and connectivity breakdown. However, cortical thinning was more evident in the neocortical regions. Cortical changes within distant regions, such as the frontal or parietal cortex, could be related to the effects of persistent changed connectivity flows, appearing as epileptiform discharges or seizures (Keller et al., 2002; Lin et al., 2007).

Our patients showed bilateral thinning of sensorimotor regions; although the reasons for these changes still need to be clarified, one possible explanation could play a role in the modified information flows between the thalamus and frontal areas, with subsequent cortico-cortical connections to the sensorimotor cortex. The fact that pre- and postcentral cortical thinning is mentioned in studies assessing cortical thickness in TLE (Bernhardt et al., 2008; Lin et al., 2007) suggests that these regions are directly involved in the focal epileptogenesis.

Thalamic volume and connectivity

The thalamus might be one of the most strongly affected regions of extratemporal gray matter in patients with focal epilepsies (Barron et al., 2013), whereas its role in seizure propagation is still unclear (Cassidy and Gale, 1998; Guye et al., 2006). In our study, thalamic volumes showed bilateral reductions and a strong negative correlation between the duration of epilepsy and volumetric values. Of particular interest is the significant relationship between temporo-thalamic cross-talk in the theta frequency band and fronto-thalamic connectivity in alpha frequency bands and the volume of both thalami.

Thalamic regions that are the most commonly associated with TLE are pulvinar, anterior nucleus, and medial dorsal nucleus (Barron et al., 2013), which have dense reciprocal connections with MTL. Volume loss in all three nuclei has been correlated with the duration and severity of medial TLE (Bernhardt et al., 2012), further supporting their conspicuous involvement in the epileptogenic network. The pulvinar part of the thalamus has several nuclear subgroups, and the most medial–dorsal part is connected to associative cortices of the frontal and temporal lobes (Rosenberg et al., 2006). Extensive and reciprocal interconnections with the cortex imply that the pulvinar serves in the facilitation of cortico-cortical transmission through thalamic loops (Saalmann and Kastner, 2011).

Local pulvinar subnetworks receiving strong extrathalamic excitatory and inhibitory inputs are more susceptible to rhythmic bursting and, respectively, to integration into remote oscillations (Fogerson and Huguenard, 2016; Wei et al., 2011). The nuclear subgroups identified in our study were medial dorsal and ventral anterior nuclei with frontal projections, and lateral and inferior nuclei of the pulvinar with temporal connections, which are in a very good agreement with existing data (Behrens et al., 2003; Bernhardt et al., 2012). Therefore, it is of utmost importance to reveal and characterize the contributions of extrathalamic-associated cortices to the initiation and generalization of hypersynchronous oscillations in thalamic nuclei (Fogerson and Huguenard, 2016).

Different nuclei of the thalamus might have specific roles in temporo-thalamic epileptogenic synchronization. The anterior nucleus of thalamus projects to superior frontal and temporal lobe structures and the bilateral stimulation of this nucleus reduce seizure frequency (Fisher et al., 2010), thus displaying an essential modulatory role in focal epileptogenesis. Quick generalization of seizures stimulated from the medial dorsal nucleus of the thalamus implies that this region has a role in the spreading of TLE seizures (Bertram et al., 2008).

Conclusions

Our combined approach of connectivity and microstructural integrity analysis of focal epileptogenic networks identified the temporal and spatial domain specificity and the frequency spectrum of synchronization leading to spikes. These networks involving the temporal and frontal cortical areas and the thalamus are hallmarks of interictal pathological activity generation utilizing mostly physiological connection but leading to a spread to adjacent regions forming the epileptogenic network. Our findings offer clear insights into the function and structure of brain networks underlying the genesis of spikes and seizures in focal epilepsy and can be possibly targeted by future therapeutic strategies.

Footnotes

Acknowledgments

The authors thank Cheryl Ernest for proofreading this article. This work was supported by a grant from the German Research Council (DFG; CRC-1193/B5 and TR-128 to S.G.). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of this article.

Author Disclosure Statement

None of the authors have any conflicts of interest to disclose. No competing financial interests exist. The authors confirm that they have read the journal's position on issues involved in ethical publication and affirm that this article is consistent with those guidelines.

References

Amjad

, Halliday

, Rosenberg

, Conway

. 1997. An extended difference of coherence test for comparing and combining several independent coherence estimates: theory and application to the study of motor units and physiological tremor. J Neurosci Methods, 73:69–79.

Barron

, Fox

, Laird

, Robinson

, Fox

. 2013. Thalamic medial dorsal nucleus atrophy in medial temporal lobe epilepsy: a VBM meta-analysis. Neuroimage Clin, 2:25–32.

Bartolomei

, Wendling

, Bellanger

J-J

, Régis

, Chauvel

. 2001. Neural networks involving the medial temporal structures in temporal lobe epilepsy. Clin Neurophysiol, 112:1746–1760.

Behrens

, Johansen-Berg

, Woolrich

, Smith

, Wheeler-Kingshott

, Boulby

, et al. 2003. Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat Neurosci, 6:750–757.

Berg

, Berkovic

, Brodie

, Buchhalter

, Cross

, van Emde Boas

, et al. 2010. Revised terminology and concepts for organization of seizures and epilepsies: report of the ILAE Commission on Classification and Terminology, 2005–2009. Epilepsia, 51:676–685.

Bernhardt

, Bernasconi

, Kim

, Bernasconi

. 2012. Mapping thalamocortical network pathology in temporal lobe epilepsy. Neurology, 78:129–136.

Bernhardt

, Worsley

, Besson

, Concha

, Lerch

, Evans

, Bernasconi

. 2008. Mapping limbic network organization in temporal lobe epilepsy using morphometric correlations: insights on the relation between mesiotemporal connectivity and cortical atrophy. Neuroimage, 42:515–524.

Bertram

, Zhang

, Williamson

. 2008. Multiple roles of midline dorsal thalamic nuclei in induction and spread of limbic seizures. Epilepsia, 49:256–268.

Bettus

, Wendling

, Guye

, Valton

, Régis

, Chauvel

, Bartolomei

. 2008. Enhanced EEG functional connectivity in mesial temporal lobe epilepsy. Epilepsy Res, 81:58–68.

10.

Cassidy

, Gale

. 1998. Mediodorsal thalamus plays a critical role in the development of limbic motor seizures. J Neurosci, 18:9002–9009.

11.

Dale

, Fischl

, Sereno

. 1999. Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage, 9:179–194.

12.

de Curtis

, Avanzini

. 2001. Interictal spikes in focal epileptogenesis. Prog Neurobiol, 63:541–567.

13.

Delorme

, Makeig

. 2004. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods, 134:9–21.

14.

A-T

, Schuff

, Kramer

, Rosen

, Gorno-Tempini

, Rankin

, et al. 2007. Different regional patterns of cortical thinning in Alzheimer's disease and frontotemporal dementia. Brain, 130:1159–1166.

15.

Faizo

, Burianová

, Gray

, Hocking

, Galloway

, Reutens

. 2014. Identification of pre-spike network in patients with mesial temporal lobe epilepsy. Front Neurol, 5:222.

16.

Fischl

, Dale

. 2000. Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad Sci U S A, 97:11050–11055.

17.

Fisher

, Salanova

, Witt

, Worth

, Henry

, Gross

, et al. 2010. Electrical stimulation of the anterior nucleus of thalamus for treatment of refractory epilepsy. Epilepsia, 51:899–908.

18.

Fogerson

, Huguenard

. 2016. Tapping the brakes: cellular and synaptic mechanisms that regulate thalamic oscillations. Neuron, 92:687–704.

19.

Fuchs

, Kastner

, Wagner

, Hawes

, Ebersole

. 2002. A standardized boundary element method volume conductor model. Clin Neurophysiol, 113:702–712.

20.

Groppa

, Herzog

, Falk

, Riedel

, Deuschl

, Volkmann

. 2014. Physiological and anatomical decomposition of subthalamic neurostimulation effects in essential tremor. Brain, 137:109–121.

21.

Groppa

, Moeller

, Siebner

, Wolff

, Riedel

, Deuschl

, et al. 2012. White matter microstructural changes of thalamocortical networks in photosensitivity and idiopathic generalized epilepsy. Epilepsia, 53:668–676.

22.

Groppa

, Muthuraman

, Otto

, Deuschl

, Siebner

, Raethjen

. 2013. Subcortical substrates of TMS induced modulation of the cortico-cortical connectivity. Brain Stimul, 6:138–146.

23.

Groppa

, Siebner

, Kurth

, Stephani

, Siniatchkin

. 2008. Abnormal response of motor cortex to photic stimulation in idiopathic generalized epilepsy. Epilepsia, 49:2022–2029.

24.

Groß

, Kujala

, Hämäläinen

, Timmermann

, Schnitzler

, Salmelin

. 2001. Dynamic imaging of coherent sources: studying neural interactions in the human brain. Proc Natl Acad Sci U S A, 98:694–699.

25.

Guye

, Régis

, Tamura

, Wendling

, Mc Gonigal

, Chauvel

, Bartolomei

. 2006. The role of corticothalamic coupling in human temporal lobe epilepsy. Brain, 129:1917–1928.

26.

Haykin

. 2001. Kalman Filtering and Neural Networks. Chichester, UK: Wiley Online Library.

27.

Holmes

, Brown

, Tucker

. 2004. Are “generalized” seizures truly generalized? Evidence of localized mesial frontal and frontopolar discharges in absence. Epilepsia, 45:1568–1579.

28.

Insausti

, Amaral

, Cowan

. 1987. The entorhinal cortex of the monkey: III. Subcortical afferents. J Comp Neurol, 264:396–408.

29.

Jones

. 2012. The Thalamus. New York: Springer Science & Business Media.

30.

Jung

T-P

, Makeig

, Humphries

, Lee

T-W

, Mckeown

, Iragui

, Sejnowski

. 2000. Removing electroencephalographic artifacts by blind source separation. Psychophysiology, 37:163–178.

31.

Jurcak

, Tsuzuki

, Dan

. 2007. 10/20, 10/10, and 10/5 systems revisited: their validity as relative head-surface-based positioning systems. Neuroimage, 34:1600–1611.

32.

Kay

, DiFrancesco

, Privitera

, Gotman

, Holland

, Szaflarski

. 2013. Reduced default mode network connectivity in treatment‐resistant idiopathic generalized epilepsy. Epilepsia, 54:461–470.

33.

Keller

, Mackay

, Barrick

, Wieshmann

, Howard

, Roberts

. 2002. Voxel-based morphometric comparison of hippocampal and extrahippocampal abnormalities in patients with left and right hippocampal atrophy. Neuroimage, 16:23–31.

34.

Ketz

, Jensen

, O'Reilly

. 2015. Thalamic pathways underlying prefrontal cortex–medial temporal lobe oscillatory interactions. Trends Neurosci, 38:3–12.

35.

Laufs

, Hamandi

, Salek-Haddadi

, Kleinschmidt

, Duncan

, Lemieux

. 2007. Temporal lobe interictal epileptic discharges affect cerebral activity in “default mode” brain regions. Hum Brain Mapp, 28:1023–1032.

36.

Lehmann

, Skrandies

. 1980. Reference-free identification of components of checkerboard-evoked multichannel potential fields. Electroencephalogr Clin Neurophysiol, 48:609–621.

37.

, Large

, et al. 2011. Using interleaved transcranial magnetic stimulation/functional magnetic resonance imaging (fMRI) and dynamic causal modeling to understand the discrete circuit specific changes of medications: lamotrigine and valproic acid changes in motor or prefrontal effective connectivity. Psychiatry Res, 194:141–148.

38.

Lin

, Salamon

, Lee

, Dutton

, Geaga

, Hayashi

, et al. 2007. Reduced neocortical thickness and complexity mapped in mesial temporal lobe epilepsy with hippocampal sclerosis. Cereb Cortex, 17:2007–2018.

39.

Lyoo

, Sung

, Dager

, Friedman

, Lee

, Kim

, et al. 2006. Regional cerebral cortical thinning in bipolar disorder. Bipolar Disord, 8:65–74.

40.

McCormick

, Contreras

. 2001. On the cellular and network bases of epileptic seizures. Annu Rev Physiol, 63:815–846.

41.

Michels

, Muthuraman

, Lüchinger

, Martin

, Anwar

, Raethjen

, et al. 2013. Developmental changes of functional and directed resting-state connectivities associated with neuronal oscillations in EEG. Neuroimage, 81:231–242.

42.

Morgan

, Sonmezturk

, Gore

, Abou-Khalil

. 2012. Lateralization of temporal lobe epilepsy using resting functional magnetic resonance imaging connectivity of hippocampal networks. Epilepsia, 53:1628–1635.

43.

Mormann

, Kreuz

, Andrzejak

, David

, Lehnertz

, Elger

. 2003. Epileptic seizures are preceded by a decrease in synchronization. Epilepsy Res, 53:173–185.

44.

Mueller

, Laxer

, Barakos

, Cheong

, Finlay

, Garcia

, et al. 2010. Involvement of the thalamocortical network in TLE with and without mesiotemporal sclerosis. Epilepsia, 51:1436–1445.

45.

Muthuraman

, Groppa

, Deuschl

. 2016. Cerebello-cortical networks in orthostatic tremor. Brain, 139:2104–2106.

46.

Muthuraman

, Hellriegel

, Hoogenboom

, Anwar

, Mideksa

, Krause

, et al. 2014. Beamformer source analysis and connectivity on concurrent EEG and MEG data during voluntary movements. PloS One, 9:e91441.

47.

Muthuraman

, Heute

, Arning

, Anwar

, Elble

, Deuschl

, Raethjen

. 2012. Oscillating central motor networks in pathological tremors and voluntary movements. What makes the difference?. Neuroimage, 60:1331–1339.

48.

Muthuraman

, Heute

, Deuschl

, Raethjen

. 2010. The central oscillatory network of essential tremor. Conf Proc IEEE Eng Med Biol Soc, 2010:154–157.

49.

Oostenveld

, Fries

, Maris

, Schoffelen

J-M

. 2011. FieldTrip: open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput Intell Neurosci, 2011:Article ID 156869.

50.

Paz

, Bryant

, Peng

, Fenno

, Yizhar

, Frankel

, et al. 2011. A new mode of corticothalamic transmission revealed in the Gria4^−/− model of absence epilepsy. Nat Neurosci, 14:1167–1173.

51.

Pittau

, Grova

, Moeller

, Dubeau

, Gotman

. 2012. Patterns of altered functional connectivity in mesial temporal lobe epilepsy. Epilepsia, 53:1013–1023.

52.

Reuter

, Schmansky

, Rosas

, Fischl

. 2012. Within-subject template estimation for unbiased longitudinal image analysis. Neuroimage, 61:1402–1418.

53.

Richardson

. 2012. Large scale brain models of epilepsy: dynamics meets connectomics. J Neurol Neurosurg Psychiatry, 83:1238–1248.

54.

Rosenberg

, Mauguière

, Demarquay

, Ryvlin

, Isnard

, Fischer

, et al. 2006. Involvement of medial pulvinar thalamic nucleus in human temporal lobe seizures. Epilepsia, 47:98–107.

55.

Rosenberg

, Amjad

, Breeze

, Brillinger

, Halliday

. 1989. The Fourier approach to the identification of functional coupling between neuronal spike trains. Prog Biophys Mol Biol, 53:1–31.

56.

Russchen

, Amaral

, Price

. 1987. The afferent input to the magnocellular division of the mediodorsal thalamic nucleus in the monkey, Macaca fascicularis. J Comp Neurol, 256:175–210.

57.

Saalmann

, Kastner

. 2011. Cognitive and perceptual functions of the visual thalamus. Neuron, 71:209–223.

58.

Siniatchkin

, Groppa

, Jerosch

, Muhle

, Kurth

, Shepherd

, et al. 2007. Spreading photoparoxysmal EEG response is associated with an abnormal cortical excitability pattern. Brain, 130:78–87.

59.

Tanaka

. 1976. Thalamic projections of the dorsomedial prefrontal cortex in the rhesus monkey (Macaca mulatta). Brain Res, 110:21–38.

60.

Tenney

, Fujiwara

, Horn

, Jacobson

, Glauser

, Rose

. 2013. Focal corticothalamic sources during generalized absence seizures: a MEG study. Epilepsy Res, 106:113–122.

61.

Van Veen

, Van Drongelen

, Yuchtman

, Suzuki

. 1997. Localization of brain electrical activity via linearly constrained minimum variance spatial filtering. IEEE Trans Biomed Eng, 44:867–880.

62.

Voets

, Beckmann

, Cole

, Hong

, Bernasconi

. 2012. Structural substrates for resting network disruption in temporal lobe epilepsy. Brain, 135:2350–2357.

63.

Wan

, Nelson

. 2001. Dual extended Kalman filter methods. In: Kalman Filtering and Neural Networks. Chichester, UK: Wiley, pp. 123–173.

64.

Wang

X-J

. 2010. Neurophysiological and computational principles of cortical rhythms in cognition. Physiol Rev, 90:1195–1268.

65.

Wei

, Bonjean

, Petry

, Sejnowski

, Bickford

. 2011. Thalamic burst firing propensity: a comparison of the dorsal lateral geniculate and pulvinar nuclei in the tree shrew. J Neurosci, 31:17287–17299.

66.

Wellmer

, Quesada

, Rothe

, Elger

, Bien

, Urbach

. 2013. Proposal for a magnetic resonance imaging protocol for the detection of epileptogenic lesions at early outpatient stages. Epilepsia, 54:1977–1987.

67.

Whitfield-Gabrieli

, Nieto-Castanon

. 2012. Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect, 2:125–141.

68.

Yaffe

, Borger

, Megevand

, Groppe

, Kramer

, Chu

, et al. 2015. Physiology of functional and effective networks in epilepsy. Clin Neurophysiol, 126:227–236.

69.

Zaveri

, Duckrow

, De Lanerolle

, Spencer

. 2001. Distinguishing subtypes of temporal lobe epilepsy with background hippocampal activity. Epilepsia, 42:725–730.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB