Localizing the Epileptogenic Zone with Novel Biomarkers

Abstract

Several noninvasive methods, such as high-density EEG or magnetoencephalography, are currently used to delineate the epileptogenic zone (EZ) during the presurgical evaluation of patients with drug resistant epilepsy (DRE). Yet, none of these methods can reliably identify the EZ by their own. In most cases a multimodal approach is needed. Challenging cases often require the implantation of intracranial electrodes, either through stereo-taxic EEG or electro-corticography. Recently, a growing body of literature introduces novel biomarkers of epilepsy that can be used for analyzing both invasive as well as noninvasive electrophysiological data. Some of these biomarkers are able to delineate the EZ with high precision, augment the presurgical evaluation, and predict the surgical outcome of patients with DRE undergoing surgery. However, the use of these epilepsy biomarkers in clinical practice is limited. Here, we summarize and discuss the latest technological advances in the presurgical neurophysiological evaluation of children with DRE with emphasis on electric and magnetic source imaging, high frequency oscillations, and functional connectivity.

Figure 1:. Overlapping cortical zones for the delineation of the EZ.

Eloquent areas mainly consist of the primary motor (green), primary somatosensory (blue), primary visual (orange), and language areas (Broca’s and Wernicke’s areas; yellow); eloquent areas are responsible for basic physiological brain functions. Irritative Zone (dotted grey area): cortical area that generates interictal spikes in the EEG. Functional deficit zone (diagonal lined grey area): cortical area that is functionally abnormal during the interictal period as it can be identified by neurological examination or neuropsychological testing. Seizure Onset Zone (red area): the brain area where seizures initiate.

Figure 2:. ESI and MSI solutions on ictal and interictal data.

A. Seizure onset (vertical red highlight area) on low-density conventional scalp EEG (selection of a −4 to 6 s time window) from a 9-year-old boy with DRE. (B) Source localization of the SOZ with Equivalent Current Dipole (ECD) of a single seizure and all seizures color-coded based on their goodness of fit (GOF) overlaid on resection (green area). C. Sensor locations and source localization of interictal spikes with ECD in a 13-year-old girl with DRE for conventional low-density (19 channels) scalp EEG, HD-EEG (72 channels), MEG (306 sensors), and iEEG (72 subdural electrodes) [11].

Figure 3:. Delineation of the irritative zone (IZ) and SOZ using ESI on iEEG data.

(A) Delineation of the IZ for a 18-year old boy underwent surgery (good outcome). Top panels show the irritative zone defined with the conventional approach by using the coordinates of most interictally active contacts (red) overlaid on patients pre-operative MRI overlaid on resection (green). Bottom panels show the irritative zone defined by ESI (cyan dipoles). (B) Delineation of the SOZ for the same patient. The figure has been obtained from [24] after permission.

Figure 4:. Spatiotemporal propagation of HFOs (ripples).

(A) Propagating ripples on filtered iEEG data (80-250 Hz) from a 10-year-old-boy with DRE. (B) Spatiotemporal propagation of ripples displayed on intraoperative photo of implanted grid of patient’s cortex and on the 3D cortical reconstruction of patient’s preoperative MRI (C) [color-coded for latency from onset (ms)]. (D) Isolated ripples (occurred either in a single channel or multi-channel) not participating on propagations [35].

Figure 5:. Virtual sensor implantation with MEG and HD-EEG for mapping noninvasively propagation of HFOs across large brain areas.

(A) Example of iEEG implantation with both subdural and depth electrodes on the left temporal and medial temporal lobe of a 12-year-old boy with DRE. (B) Virtual channel placement based on the coordinates of iEEG electrodes (matched locations). Ripple propagation on MEG (C) and HD-EEG (D) virtual sensors across time [61].

Figure 6:. FC estimates from iEEG.

Pairwise FC matrices (right) are estimated from iEEG. The connectivity matrix for a brain network comprising N channels is a N×N matrix: diagonal elements of the matrix represent the connectivity of each node with itself (therefore we will assume that they contain only zero values); off-diagonal elements of the connectivity matrix represent the connectivity between pairs of distinct channels.

Figure 7:. Patient-level strength selectivity analysis based on iEEG data.

Spatial maps of FC (nodal strength in beta band) (left) along with corresponding 2D heat maps of nodal strength in all frequency maps (right) for a patient with good (a) and poor (b) surgical outcome. Resection zones are highlighted in green. The figure has been obtained from [71] after permission.

Figure 8:. Novel technological improvements in mapping the EZ.

Pediatric HD-EEG recordings (a) that can be performed simultaneously with MEG (b) at Cook Children’s Medical Center (Fort Worth, TX, USA). Permission has been obtained from the parents of the child to use his image for the purposes and dissemination of knowledge.