Authors :
Presenting Author: Ionut-Flavius Bratu, MD – Timone Hospital, Marseille, France
Samuel Medina Villalon, Eng – Timone Hospital, Marseille, France; Elodie Garnier, Eng – Aix-Marseille University, Marseille, France; Christian Benar, PhD, Eng – Team lead DynaMap, Systems Neuroscience Institute, Aix-Marseille University, Marseille, France; Fabrice Bartolomei, MD, PhD, Prof – Head of Department, Clinical Neurophisiology, Timone Hospital, Marseille, France
Rationale:
Even though clinical recovery may be achieved shortly after the electrical or clinical termination of a seizure, electroencephalography alterations may persist well after. In this study, we quantified the post-ictal stereo-electroencephalography (SEEG) signal complexity changes using permutation entropy (PE) in light of the epileptogenic zone network's (EZN) location and lateralization.
Methods:
PE was performed on ictal and peri-ictal SEEG recordings of 120 patients with drug-resistant temporal lobe epilepsy who had undergone invasive evaluation at La Timone Hospital, Marseille. PE analysis was done using AnyWave software (https://meg.univ-amu.fr/wiki/AnyWave) and an in-house plug-in. We used an embedding dimension = 3, lag = 1 and time window length = 5 seconds with an overlap of 2.5 seconds. Global PE values were computed from all non-artifacted channels during preictal, postictal and ictal periods of spontaneous seizures. The mean preictal baseline global complexity was established within two standard deviations (SD). It was considered that the SEEG signal returned to its preictal complexity ("come-back point") when for 25 consecutive seconds onward (10 x 2.5 seconds overlap) the mean global PE value was again in the 2SD interval. The period between the electrical seizure termination and the "come-back point" was named post-ictal alteration time (PAT). The patients were divided based on their SEEG-defined EZN in uni-temporal (lateral LT, mesial MT and lateral-mesial LMT) or bi-temporal.
Results:
We analyzed 95 patients with uni-temporal EZN (12 LT, 69 MT and 14 LMT) and 25 patients with bi-temporal EZN. PE was computed on 211 seizures: 19 from LT patients (4 left EZN and 15 right), 117 from MT patients (62 left EZN and 55 right), 26 from LMT patients (14 left EZN and 12 right), and 49 seizures from bi-temporal EZN patients. The longest PAT was found in left LMT: 3745.5 seconds (approx. 1 hour and 2 minutes) (maximum), 629.28 (mean) ± 934.85 seconds (SD). In the right-hemisphere EZN group, the longest PAT was observed in the MT group and it reached 3390 seconds (maximum), 598.59 (mean) ± 757.58 seconds (SD). PAT values were higher in left-hemisphere EZN seizures in the LTM (p 0.022) and LT (p 0.009) groups. Age at epilepsy onset was positively correlated with the maximal and average values per patient. PAT tended to be lower in Engel I patients compared to Engel non-I patients (p 0.08).
Conclusions:
This study quantified for the first time the alterations in signal complexity and their dynamics in the postictal period. It paves the way for a better understanding of this state and its impact on clinical issues (e.g., semiology, cognition, postictal deficit).
Funding: No funding.