States of functional connectivity flow and their multiplex dynamics in human epilepsy and postictal aphasia

The mechanisms that cause aphasia as a transient post-seizure symptom in epileptic patients are yet unknown. We analyse intracranial EEG (sEEG) recordings of patients suffering from pharmaco-resistant epilepsy with postictal aphasia. We study the Functional Connectivity (FC) between different cortical sites in a time- and frequency-resolved manner, representing each recording as a time-varying, multilayer network (dynamic multiplex). We studied in particular: the rate of overall reconfiguration of links from one frame to the next, or dynamic Functional Connectivity (dFC) speed; and the stability of network modules through time, by means of a dynamic modular Allegiance (dA) analysis. The combination of these two approaches allows identifying states of “Functional Connectivity flow” (beyond connectivity states), defined as epochs in which network reconfiguration occurs with comparable speed and degree of spatio-temporal coordination. Our unsupervised analyses reveal then that high-frequency dFC is slowed down in a long postictal phase lasting well beyond the ictal episodes themselves. Furthermore, a pathological state of slow and poorly structured network flow consistently co-occurs with episodes of aphasia symptoms annotated by the clinicians. In conclusion, our multiplex network dynamics description cast light on functional mechanisms of postictal cognitive dysfunction at the level of individual patients.