Combining Gamma With Alpha and Beta Power Modulation for Enhanced Cortical Mapping in Patients With Focal Epilepsy

Abstract

About one third of patients with epilepsy have seizures refractory to the medical treatment. Electrical stimulation mapping (ESM) is the gold standard for the identification of "eloquent" areas prior to resection of epileptogenic tissue. However, it is time-consuming and may cause undesired side effects. Broadband gamma activity (55-200 Hz) recorded with extraoperative electrocorticography (ECoG) during cognitive tasks may be an alternative to ESM but until now has not proven of definitive clinical value. Considering their role in cognition, the alpha (8-12 Hz) and beta (15-25 Hz) bands could further improve the identification of eloquent cortex. We compared gamma, alpha and beta activity, and their combinations for the identification of eloquent cortical areas defined by ESM. Ten patients with intractable focal epilepsy (age: 35.9 ± 9.1 years, range: 22-48, 8 females, 9 right handed) participated in a delayed-match-to-sample task, where syllable sounds were compared to visually presented letters. We used a generalized linear model (GLM) approach to find the optimal weighting of each band for predicting ESM-defined categories and estimated the diagnostic ability by calculating the area under the receiver operating characteristic (ROC) curve. Gamma activity increased more in eloquent than in non-eloquent areas, whereas alpha and beta power decreased more in eloquent areas. Diagnostic ability of each band was close to 0.7 for all bands but depended on multiple factors including the time period of the cognitive task, the location of the electrodes and the patient's degree of attention to the stimulus. We show that diagnostic ability can be increased by 3-5% by combining gamma and alpha and by 7.5-11% when gamma and beta were combined. We then show how ECoG power modulation from cognitive testing can be used to map the probability of eloquence in individual patients and how this probability map can be used in clinical settings to optimize ESM planning. We conclude that the combination of gamma and beta power modulation during cognitive testing can contribute to the identification of eloquent areas prior to ESM in patients with refractory focal epilepsy.

Keywords: alpha frequency band; beta frequency band; broadband gamma frequency; drug-resistant epilepsy; electrical stimulation mapping; electrocorticography-based functional mapping; epilepsy surgery; frequency-based cortical mapping.

FIGURE 1

ESM categories and ECoG responses. (A) Electrodes tested during electrical-cortical stimulation mapping (ESM) in 10 patients projected in the common space, color coded according to the ESM category. LH, left hemisphere; RH, right hemisphere; Ventral, ventral (caudal) view of left and right hemispheres; Dorsal, dorsal (cranial) view of left and right hemispheres. (B) Time frequency representations (TFR) of time periods of the delayed match-to-sample task from all tested electrodes grouped by the ESM category. The TFRs represent the three different trial events (i.e., sound onset, letter onset, and button-press), which are illustrated by the three vertical dotted lines in each TFR). (C) Illustration of the time course in the task (upper cartoon, events are represented at their average time after sound onset) and the “exaggerated” time between the events used for display time (lower cartoon).

FIGURE 2

AUROC analysis. (A) Normalized change in ECoG signal power spectral density from baseline from eloquent (red) and non-eloquent electrodes (blue). Shading shows standard error, center line shows mean. (B) AUROC values across frequencies, pale green, non-significant, dark green significant. Black lines show 2.5 and 97.5 percentiles from permutation distribution. Dotted line shows chance performance. Background shading indicates filter boundaries for each band. (C) Gamma power modulation during each task event (icons indicate sound onset, letter onset, and button-press). (D) AUROC values across time for gamma. (E,F) Alpha power and AUROC values, as in (C,D). (G,H) Beta power and AUROC values, as in (C,D).

FIGURE 3

Temporal AUROC and model comparison. (A) AUROC values using GLM approach for gamma, alpha, and beta (solid lines), and their combinations (dashed lines). (B) Distribution of the AUROC values from 1000 bootstrap samples for single models. Vertical lines show AUROC values without resampling. (C) Pairwise comparison from bootstrap distribution of alpha (red) and beta (blue) AUROC values against gamma. Vertical lines show mean percentage change from gamma-only model. (D,E) Distribution of AUROC values for combination models, as in (B). (E) Pairwise comparisons of combination models with gamma-only model.

FIGURE 4

AUROC analysis per functional category. (A–C) Gamma, Alpha and Beta power modulation and SE (shades) for three subtypes of electrodes (input, output and non-eloquent) during the three events of the task. (D) Distribution of the AUROC values from 1000 bootstrap samples for single models and input channels. (E) Pairwise comparison from bootstrap distribution of alpha (red) and beta (blue) AUROC values against gamma for input channels. (F,G) Same as (D,E) for output channels. (H–K), same as (D–G) using combined models.

FIGURE 5

Influence of attention on AUROC values. (A–C) Gamma, Alpha and Beta power across time for eloquent (red) and non-eloquent (blue) electrodes during the active (DMTS, solid lines, darker shading-SE) and the passive (dashed lines, lighter shading-SE) tasks. (D) Effect of attention on diagnostic ability per model. Vertical lines show 5 and 95 percentiles from 1000 bootstrap samples, horizontal offset to aid visibility. (E) Pairwise comparison between active and passive tasks per single band models. Negative values imply a better performance in the active task. (F) Same as (E) but using combination models.

FIGURE 6

Probabilistic maps of eloquence. (A) Relationship between GLM response (i.e., prediction of the binomial probability that an electrode is eloquent) and probability of eloquence (leftward axis) and numbers of electrodes (rightward axis). Dot colors indicate color-scale used in other panels. (B) Electrode locations projected onto the individual patient MRI in five patients. Dot color indicates GLM response, dot centers indicate ESM results, white: non-eloquent, black: eloquent. Red rings; seizure onset zones. Bodies in the lower left of each panel indicate the brain orientation. Values above each brain show the quantification of the performance of the GLM. This is given in terms of the mean and standard deviation GLM response of ESM positive (eloquent) and negative (non-eloquent) electrodes, and an AUROC value where patients have both positive and negative electrodes. Non-eloq, non-eloquent; mn, mean; SD, standard deviation; eloq, eloquent; AUROC, area under the receiver operating characteristic curve.