Electrical brain-stimulation paradigm for estimating the seizure onset site and the time to ictal transition in temporal lobe epilepsy

https://doi.org/10.1016/j.clinph.2004.08.021Get rights and content

Abstract

Objective

To explore and validate a novel stimulation and analysis paradigm proposed to monitor spatial distribution and temporal changes of the excitability state in patients with temporal lobe epilepsy (TLE).

Methods

We use intermittent pulse stimulation in the frequency range 10–20 Hz. A quantitative measure of spectral phase de-modulation, the relative phase clustering index (rPCI) was applied to the evoked EEG signals, measured from electrodes implanted in the hippocampal formation.

Results

We found that in the interictal periods, high values of rPCI recorded from specific sites were correlated with the most probable seizure onset sites (SOS). Furthermore we found that high values of rPCI from certain locations correlated with shorter time intervals to the next seizure.

Conclusions

Our clinical findings indicate that although the precise moment of ictal transitions is in general unpredictable, it may be possible to estimate the probability of occurrence of some epileptic seizures.

Significance

The use of the rPCI for probabilistic forecasting of upcoming epileptic seizures is warranted. rPCI measurements may be used to guide interventions with the aim of modifying local tissue excitability that ultimately might prevent ictal transitions.

Introduction

Epilepsy is a disease that features rapid or gradual transitions from a ‘normal’ functional state of the brain to an ‘epileptic’ state. These transitions in the majority of cases are not related to a particular known factor and are not governed by a regular time schedule. We refer therefore to the epileptic disorder as a ‘dynamical disease’ (Lopes da Silva et al., 2003). There has been a long-standing discussion in the literature (de Curtis and Avanzini, 2001) on the issue of relating interictal neuronal activity and any sort of forecasts concerning time, location and type of epileptic activity in patients and experimental animal models. The challenge of predicting the exact time and/or location of an epileptic seizure onset has been taken by many researchers in the recent years. The literature and variety of methods (Iasemidis, 2003, Lehnertz et al., 2003, le Van Quyen et al., 2000, Mormann et al., 2003, Worrell et al., 2004) is growing, but to our best knowledge no single technique has been validated to give definite answer to those problems. Some of the claimed results have been challenged (Aschenbrenner-Scheibe et al., 2003, Lai et al., 2003) which makes the issue even more controversial. Most studies with this objective have been carried out using spontaneous EEG signals, although Wilson et al. (1998) have used field evoked potentials as a measure of the excitability state of local neuronal networks in the limbic system. We assumed that it would be possible to obtain relevant information about the dynamical state of these networks through the analysis of activities evoked by local stimulation with pulse trains. We searched for the information in the phase domain of these evoked activities, inspired by our finding in photosensitive patients where the enhancement of phase clustering (rPCI) of some frequency components was shown to anticipate the transition to an epileptic seizure during intermittent light stimulation (Kalitzin et al., 2002, Parra et al., 2003).

We first attempt to identify the site of ictal onset in patients with TLE, undergoing invasive presurgical diagnostic evaluation. We use a particular measure of phase de-modulation, the rPCI, measured interictally during intermittent electrical stimulation. The second objective is to explore the possibility and to present a ‘proof-of-principle’ for this methodology with respect to seizure prediction. We should note that the question of seizure prediction suffers from a fundamental imprecision: what does one precisely mean by saying that an ictal transition can be predicted? We propose here that this imprecision can be removed by using a probabilistic formulation. This means that an operational definition of seizure anticipation should be based on the probability distribution of a given measure related to the distribution of time intervals between the moment of estimation of this measure and the moment of seizure onset. We can assess the predictive value of such distribution and therefore define quantitatively seizure ‘predictability’. According to this concept, predictability does not translate directly into a well-defined time of forecasting an ictal transition, it rather consists of an estimate of the chance of a seizure occurring during a certain period in the future relative to a particular moment of observation.

Section snippets

Patients and implantation

All 6 patients participating in this study were candidates for resective epilepsy surgery who underwent phase 2 invasive EEG monitoring because of lack of congruence of non-invasive presurgical evaluation studies (Dutch Collaborative Epilepsy Surgery Program, 1993). No other selection criterion applied and all patients implanted from 2001 onwards participated in the present study. Implantation was carried out through two frontal parasagittal circular craniotomies 3 cm in diameter in case of

Interictal lateralization and localization

Inter-ictal localization and lateralization results are summarised in columns 1–5 of Table 1. The lateralization for all patients except patient 3 who was implanted only unilaterally, was done on the basis of the rPCI values collected interictally (>24 h before first seizure) from all left and right hippocampal contacts. The statistical test shows higher rPCI values measured from the SOS containing hemisphere in 4 out of 4 available cases (patients 1, 2, 4 and 5). In one case (patient 6) there

Discussion

Theoretically, there is no principal argument imposing a deterministic relation between the brain's dynamics in its ‘normal’ state and the possibilities and features of an ‘epileptic’ state. Our previous studies (Lopes da Silva et al., 2003) have shown that bistable systems can exhibit both types of behaviour and switches between them can be due to noisy fluctuations. Such analysis, however, does not exclude indirect, possibly non-causal relations between interictal observable features and

References (23)

  • S. Kalitzin et al.

    Enhancement of phase clustering in the EEG/MEG gamma frequency band anticipates transition to paroxysmal epileptiform activity in epileptic patients with known visual sensitivity

    IEEE Trans Biomed Eng

    (2002)
  • Cited by (106)

    • Clinical utility of EEG in diagnosing and monitoring epilepsy in adults

      2018, Clinical Neurophysiology
      Citation Excerpt :

      The highly individual nature of patient seizure patterns is another striking feature, suggesting that algorithms will need to be individually optimized. Seizure prediction techniques are being pursued and hold promise for notifying PWE in advance of a seizure (Kalitzin et al., 2005; Stypulkowski et al., 2013, 2014; Freestone et al., 2013), but are in the early stages of development and suffer from limitations related to cost and invasiveness. Outpatient aEEG monitoring is 51–65% less expensive than inpatient 24-h VEM and less restrictive to patients, allows patient assessment in their natural environment with exposure to daily natural seizure triggers (Schomer, 2006).

    View all citing articles on Scopus
    View full text