Abstracts

Rationale: Mesial temporal lobe epilepsy (MTLE) is the most common and most commonly medically refractory form of epilepsy. This disease is characterized by epileptic discharges emanating from the hippocampus with patients typically exhibiting cognitive deficits related to its dysfunction. However, structural imaging often does not reveal pathology within the hippocampus. In contrast, functional imaging such as magnetoencephalography (MEG) may be sensitive to these altered functional networks. We utilized MEG in a novel method to investigate disturbances in the functional organization of the hippocampus in MTLE using connectivity in theta rhythm.Methods: Four healthy controls and 4 patients with lateralized MTLE underwent 5 minutes of non-task, eyes-open recordings using a 306-sensor VectorView Elekta MEG. Bad channels and time segments were excluded from analysis. Signals were notch filtered at 60Hz and its harmonics and then bandpass filtered between 1-240Hz. Heartbeats and eye blinks were isolated and removed using independent component analysis. The deep brain activity model implemented in Brainstorm toolbox was used to model the source activity to better account for the subcortical sources. The source signals were bandpass filtered into the theta band (4-8Hz) and transformed using the Hilbert transform. Functional connectivity was estimated using the weighted phase lag index (WPLI) by averaging across 30 randomly selected, 4-second epochs. The eta2 coefficient between pairwise hippocampal sources in each hippocampus was calculated as a similarity index between that pair of sources. The values were squared to accentuate the relationship to dissimilar sources.The hippocampal eta2 coefficient values were used to estimate the number of clusters in each hippocampus using the Calinski-Harabasz criterion in a range between 2 and 20 clusters. Finally, each hippocampus was clustered using the K-means algorithm into the optimal number of clusters based on the eta2 coefficient values. The number of clusters generated across groups was tested for significance using a Kruskal-Wallis analysis of variance.Results: In the MTLE group, hippocampi were optimally clustered into a greater number of clusters compared to healthy controls (mean ranks 6.25 vs 2.75, p=0.03). MTLE subjects demonstrated spatially disorganized clusters, in sharp contrast to healthy subjects in which a symmetrical spatial distribution of clusters along the anterior-posterior axis was observed. This increased heterogeneity is consistent with altered functional organization of the hippocampus in patients with MTLE.Conclusions: This is the first study to demonstrate functional subdivisions of the hippocampus using MEG. These novel data warrant further exploration of the utility of this method as a possible predictor of hippocampal cognitive function or seizure focus location.