Abstracts

Rationale: Accurately delineating the epileptogenic network is crucial to planning surgery for drug-resistant focal epilepsy. Many patients with discordant or inconclusive imaging findings undergo intracranial EEG (iEEG) monitoring to define the seizure onset zone (SOZ). We hypothesize that using a normative iEEG atlas to benchmark deviations from normal brain dynamics provides a data-driven method to identify surgical targets.

Methods: We constructed a normative iEEG atlas by augmenting a 106-subject normative iEEG atlas (Frauscher et al., Brain, volume 141, 2018, pages 1130-1144) from the Montreal Neurological Institute (MNI) with 60 subjects carefully selected from the Hospital of Pennsylvania (HUP). This rigorously harmonized, clinically validated MNI-HUP normative iEEG atlas aggregates 2360 electrodes placed in brain areas deemed clinically normal and located outside of epileptogenic tissue. We map each electrode to a predefined parcellated region of interest (ROI) and compute normative distributions of spectral power, wavelet entropy, correlation, and coherence in each ROI. To demonstrate clinical utility, we selected a separate set of 2577 abnormal iEEG channels implanted in irritative or seizure onset areas across 60 patients. We quantitatively compared epileptic iEEG channels to normative data and mapped patient-specific abnormalities. Figure 1 provides an overview of our methods.

Results: We show that normative iEEG atlas mapping quantifies abnormalities that are not obvious from electrographic recordings (Figure 2A,B). The multivariate combination of spectral power, wavelet entropy, correlation, and coherence in a random-forest model quantified iEEG abnormalities better than any of these measures individually, and we identified SOZ channels with an area under receiver operating characteristic curve (AUC) of 81% (Figure 2C). We also find that patients with lesional MRI have a greater level of abnormality in the SOZ compared to non-lesional patients (Figure 2D). For seizure onset zones within the amygdala and hippocampus, the most routinely ablated areas in treated patients, measures of connectivity abnormality are more enhanced than univariate measures of abnormal neural activity (Figure 2E).

Conclusions: This study establishes a framework and prospective, data-driven method to guide epilepsy surgery by aggregating iEEG studies. We believe that expanding this atlas with data from a large number of epilepsy centers, comparing individual patients to both normative data and matching abnormal patterns, will greatly improve this method and patient outcomes over time. We publicly share our atlas so that others may build upon the data and methods which we present here.

Funding: Please list any funding that was received in support of this abstract.: J.B. acknowledges funding from NIH 6T32NS091006. B.L. acknowledges funding from the Pennsylvania Tobacco Fund, NIH R01NS099348 and DP1: NS122038, the Mirowski Family Foundation, Jonathan Rothberg, and Neil and Barbara Smit.