Exploratory analysis of epileptic and non-epileptiform hippocampal network organization across spatial and frequency domains

Understanding how epileptic networks may disrupt information processing requires examining the spatial organization of functional connectivity. Here, we present an exploratory case-series analysis of stereotactic EEG (SEEG) recordings from seven hippocampi of epileptic patients, six exhibiting epileptiform activity and one outside the epileptogenic zone. Phase coherence across electrode pairs was evaluated within four-node directed pathways as a function of inter-electrode distance to characterize the spatial-frequency structure of functional interactions. We introduce a novel framework that decomposes phase coherence into two complementary components-propagation and interaction phases-capturing both global coupling strength and its spatial gradient. We show that this framework yields meaningful and consistent results only when applied to electrode pairs embedded within directed pathways, demonstrating that network organization constrains the spatial structure of coherence and that pathway-based selection is essential to reveal this structure. Within this constrained dataset, we observed variability in coupling strength and spatial gradients across hippocampi. Specifically, the hippocampus not involved in the epileptogenic network exhibited lower global coupling strength and steeper spatial gradients, consistent with more modular and spatially differentiated communication. In contrast, hippocampi with epileptiform activity showed higher global coupling and flatter spatial gradients, suggesting reduced differentiation between local and remote interactions. Simulations using a one-dimensional Kuramoto oscillator model reproduced this pattern, indicating that increased coupling strength can lead to a collapse of spatial gradients. These findings illustrate how spatial coherence gradients and pathway-based analysis provide a useful framework for future studies aimed at validating network alterations in epilepsy using adequately powered cohorts.