Amplitude entropy captures chimera-like behavior in epileptic seizure dynamics

Epilepsy is a neurological disease characterized by epileptic seizures, which manifest with localized high-synchrony, high-amplitude activity that spreads from an onset zone to the rest of the epileptic network. Chimeras, defined as states of co-occurring synchrony and asynchrony in symmetrically coupled networks are increasingly invoked for characterization of seizures. In particular, chimera-like states have been observed during the transition from a normal (asynchronous) to a seizure (synchronous) network state. However, chimeras in epilepsy have only been investigated with respect to the varying phases of oscillators. We propose a novel method capturing the characteristic pronounced changes in the recorded EEG amplitude during seizures by estimating chimera-like states directly from the signals in a frequency-and time-resolved manner. We test the method on a publicly available intracranial EEG dataset of 16 patients with focal epilepsy. We show that the proposed measure, titled Amplitude Entropy, is sensitive to seizure onset dynamics, demonstrating its significant increases during seizure as compared to before and after seizure. This finding is robust across patients, their seizures, and different frequency bands. In the future, Amplitude Entropy could serve as a tool for seizure detection, but also help to characterize amplitude chimeras in other networked systems with characteristic amplitude dynamics.