The “Connectivity Epileptogenicity Index ” (cEI), a method for mapping the different seizure onset patterns in StereoElectroEncephalography recorded seizures

Highlights

• Connectivity epileptogenicity index (cEI) is a method to quantify seizure onset.

• cEI combines a directed connectivity measure (“out-degrees”) and the original epileptogenicity index (EI).

• For seizures with slow onset patterns, cEI gave a better estimation than EI.

Abstract

Objective

Localization of epileptogenic brain regions is a crucial aim of pre-surgical evaluation of patients with drug-resistant epilepsy. Several methods have been proposed to identify the seizure onset zone, particularly based on the detection of fast activity. Most of these methods are inefficient to detect slower patterns of onset that account for 20–30% of commonly observed Stereo-Electro-Encephalography (SEEG) patterns. We seek to evaluate the performance of a new quantified measure called the Connectivity Epileptogenicity Index (cEI) in various types of seizure onset patterns.

Methods

We studied SEEG recorded seizures from 51 patients, suffering from focal drug-resistant epilepsy. The cEI combines a directed connectivity measure (“out-degrees”) and the original epileptogenicity index (EI). Quantified results (Out-degrees, cEI and EI) were compared to visually defined seizure onset zone (vSOZ). We computed recall (sensitivity) and precision (proportion of correct detections within all detections) with vSOZ as a reference. The quality of the detector was quantified by the area under the precision-recall curve.

Results

Best results (in terms of match with vSOZ) were obtained for cEI. For seizures with fast onset patterns, cEI and EI gave comparable results. For seizures with slow onset patterns, cEI gave a better estimation of the vSOZ than EI.

Conclusions

We observed that cEI discloses better performance than EI when seizures starts with slower patterns and equal to EI in seizures with fast onset patterns.

Significance

The cEI is a promising new tool for epileptologists, that helps characterizing the seizure onset zone in sEEG, in a robust way despite variations in seizure onset patterns.

Introduction

Epilepsy surgery can significantly improve the quality of life of patients with drug-resistant epilepsy. It consists in the removal of the brain regions generating seizures. These regions, (called the “Epileptogenic zone ” (Bancaud, 1980)) must be thus precisely defined from anatomo-electro-clinical observations obtained during pre-surgical evaluation. In a large number of cases, this evaluation involves invasive EEG recordings, in particular stereo-electroencephalography (SEEG) (Bancaud and Talairach, 1965, Cardinale et al., 2016, Isnard et al., 2018).

Defining the epileptogenic zone from SEEG can be challenging (Bartolomei et al., 2017, Bartolomei et al., 2018). The epileptogenic zone is increasingly recognized as a network of hyperexcitable connected regions generating seizures and secondary leading to ictal spreading in propagation networks (Bartolomei et al., 2017). The electrical onset of seizures may appear quite complex in its aspect, both in term of involved frequencies and in term of spatial extension. In 75% of cases, the seizure onset patterns involve high frequencies, most often in the high beta or gamma band (Perucca et al., 2014, Lagarde et al., 2019).

In this context, methods for quantifying the epileptogenic zone have been developed over the past ten years in order to complement the SEEG interpretation (for review see (Andrzejak et al., 2015, Bartolomei et al., 2017)). These are based on a spectral analysis of the SEEG signal obtained in each region, using energy in the high frequency (beta-gamma bands), potentially combined with other measures (low frequency deactivation, preictal activity). The first method that was developed, and the one with the largest patient population published until now, is the Epileptogenicity Index (EI) (Bartolomei et al., 2008). EI combines the ratio of fast frequencies (beta/gamma) relative to slower frequencies and the time of involvement of each region. This method inspired a number of studies aimed at identifying the epileptogenic zone from SEEG recordings (David et al., 2011, Gnatkovsky et al., 2011, Gnatkovsky et al., 2014, Grinenko et al., 2018). However, as most of these methods are based on the detection/mapping of high frequencies, they are inefficient to detect slower patterns of onset that account for 20–30% of commonly observed SEEG patterns (Singh et al., 2015, Lagarde et al., 2019). This is a clear a limitation since seizures with slow onset have worse prognosis for surgery outcome than seizures with fast activity at onset (Singh et al., 2015, Lagarde et al., 2019). It has been suggested that seizures with slower patterns are more distributed and that these slow patterns are related to etiology - in particular type 1 focal cortical dysplasia (Lagarde et al., 2019). It should be noted that the EI cannot be modified to measure low-frequency onsets because these serve as the denominator of the energy ratio from which it is calculated.

In this context, other methods based on functional connectivity have been proposed to study focal seizures onset from intracranial recordings (Brazier, 1969, Lopes da Silva et al., 1989, Gotman and Levtova, 1996, Bartolomei et al., 2004, van Mierlo et al., 2013, Courtens et al., 2016). These approaches measure linear or nonlinear properties of the relationships between SEEG signals, during the seizure onset period or the transition from interictal to ictal periods. Thus, the functional connectivity between SEEG signals tends to be maximal before the emergence of the fast discharges and to decrease just after, followed by later increase during the seizure course (Courtens et al., 2016). Some studies have used measure of causality (or directed or effective connectivity) and have shown that the definition of ”leader/driver” regions may add important clues into the definition of the network generating seizures (Bartolomei et al., 2004, Varotto et al., 2012, van Mierlo et al., 2013, Courtens et al., 2016).

The epileptogenicity of a brain network (i.e. its ability to generate seizures) is related to the level of excitability and the level of connectivity (Proix et al., 2014, Hebbink et al., 2017). In the present study we therefore propose an approach which combines two types of measures in the same quantity. A measure of connectivity and a measure of the capacity of regions to generate seizures with a short delay (EI). The objective is to improve sensitivity while better accounting for the diversity of seizure onset patterns, including slower ones. The connectivity measure was based on the nonlinear correlation coefficient (h²) and on graph theory measures (out-degrees) (Courtens et al., 2016). We focused our study on a cohort of patients in whom the seizure onset patterns (SOP) have been previously defined (Lagarde et al., 2016), relatively homogeneous in terms of etiology (malformation of cortical development) and included a proportion of slow onset patterns (20%).