Automated diagnosis of temporal lobe epilepsy in the absence of interictal spikes.

Thibault Verhoeven,Christoph Michel,Serge Vulliemoz,Roland Wiest,Gijs Plomp,Pieter Van Mierlo,Margitta Seeck,Aljoscha Thomschewski,Francesca Pittau,Joni Dambre,Ana Coito,Eugen Trinka,Karl Schaller

doi:10.1016/j.nicl.2017.09.021

Abstract

ObjectiveTo diagnose and lateralise temporal lobe epilepsy (TLE) by building a classification system that uses directed functional connectivity patterns estimated during EEG periods without visible pathological activity.MethodsResting-state high-density EEG recording data from 20 left TLE patients, 20 right TLE patients and 35 healthy controls was used. Epochs without interictal spikes were selected. The cortical source activity was obtained for 82 regions of interest and whole-brain directed functional connectivity was estimated in the theta, alpha and beta frequency bands. These connectivity values were then used to build a classification system based on two two-class Random Forests classifiers: TLE vs healthy controls and left vs right TLE. Feature selection and classifier training were done in a leave-one-out procedure to compute the mean classification accuracy.ResultsThe diagnosis and lateralization classifiers achieved a high accuracy (90.7% and 90.0% respectively), sensitivity (95.0% and 90.0% respectively) and specificity (85.7% and 90.0% respectively). The most important features for diagnosis were the outflows from left and right medial temporal lobe, and for lateralization the right anterior cingulate cortex. The interaction between features was important to achieve correct classification.SignificanceThis is the first study to automatically diagnose and lateralise TLE based on EEG. The high accuracy achieved demonstrates the potential of directed functional connectivity estimated from EEG periods without visible pathological activity for helping in the diagnosis and lateralization of TLE.

Highlights

Mesial temporal lobe epilepsy (TLE) is the most common type of pharmaco-resistant epilepsy in adults
Patients were retrospectively selected from the highdensity EEG database of the University Hospital of Geneva, University Hospital of Bern and Paracelsus Medical University in Salzburg according to the following inclusion criteria: drug-resistant TLE, unilateral anteromedial localization of the epileptogenic zone confirmed by good surgical outcome (Engel's class I or II, after at least 12 months post-operative follow-up), intracranial EEG or concordant presurgical evaluation methods and the existence of at least a 10–15 min restingstate eyes-closed high-density EEG recording (96–256 channels)
The automatic diagnosis of TLE based on EEG periods without Interictal Epileptiform Discharges (IEDs) has several important advantages: (Laufs, 2012) resting-state EEG can be recorded in less than 1 h, overcoming long term monitoring and its related costs, no IEDs or ictal activity are required, enabling the use of this method in patients with low seizure and/or IEDs frequency, (Wilke et al, 2009) the features that result in the best classification provide insight into the differences between the groups and the mechanism of action of TLE

Summary

Introduction

Mesial temporal lobe epilepsy (TLE) is the most common type of pharmaco-resistant epilepsy in adults. In order to estimate the localisation of the epileptogenic zone, Electroencephalography (EEG) is recorded to identify pathological activity such as seizures or Interictal Epileptiform Discharges (IEDs). Several studies have shown that directed functional connectivity measures based on intracranial EEG can help to identify the irritative zone and the seizure onset zone (van Mierlo et al, 2013; Wilke et al, 2009; van Mierlo et al, 2014). Directed functional connectivity applied to brain sources estimated from high-density scalp EEG revealed interictal network patterns concordant with cognitive deficits in TLE (Coito et al, 2015) and significant connectivity differences in TLE compared to healthy controls in the absence of interictal spikes (Coito et al, 2016)

Methods

Results

Discussion

Conclusion