Authors :
Presenting Author: Brian Ervin, PhD – Cincinnati Children's Hospital Medical Center
Ravindra Arya, MD, DM – Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine, University of Cincinnati College of Engineering and Applied Science; Jason Buroker, BS – Neurology – Cincinnati Children’s Hospital Medical Center; Hansel Greiner, MD – Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine; Katherine Holland, MD – Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine; Paul Horn, PhD – Neurology – Cincinnati Children’s Hospital Medical Center; James Leach, MD – Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine; Francesco Mangano, DO – Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine; leonid Rozhkov, MS – Neurology – Cincinnati Children’s Hospital Medical Center; Craig Scholle, BS – Neurology – Cincinnati Children’s Hospital Medical Center; Jesse Skoch, MD – Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine
Rationale:
Intracranial language mapping is an essential part of epilepsy surgery planning. Task-related high gamma power modulation (HGM) in intracranial electroencephalography (iEEG) has been identified as a safe, accurate alternative to the clinical standard of electrical stimulation mapping. However, common language tasks require sustained cooperation from the patient, and do not engage all cognitive subdomains of the language network. Here, we present Behavior-EEG Spectral Power Correlation (BESPoC), a novel method for iEEG functional mapping, which does not rely on trial-averaging and thus does not require repetitive tasks. Instead, it enables mapping of naturalistic behavior, such as conversation, which is more familiar for the patient, more ecologically valid, and could be especially useful for pediatric patients or patients with cognitive delays. We hypothesize that BESPoC can identify cognitive substrates underlying expressive and receptive language processing from spontaneous conversation.
Methods:
For conventional HGM analysis, trials were aligned on test condition onset, bandpass filtered 50-150 Hz, and Hilbert transformed. The log-transformed power was averaged across trials, and the power during test was compared to rest via Welch’s t-test. For BESPoC analysis, iEEG data of the duration of the task recording was wavelet-transformed 50-150 Hz. Pearson’s correlation measured the strength of the linear relationship between high gamma power and behavioral channels (patient microphone, story playback). A distribution of correlation coefficients was obtained via permutation and tested for statistical significance. Generalized linear models were fitted to predict sites of significant conventional HGM from the correlation p-values. Conversation analysis was performed via BESPoC on both patient and partner microphone channels representing expressive and receptive language, respectively.
Results:
Ninety one patients (
M55
F36) ranging from 2 to 29 years (12.5 ± 5.5) were implanted in left hemisphere, right hemisphere, and bilaterally (32, 28, 31) with 4 to 22 electrode arrays (13.3 ± 3.4) having 42 to 253 iEEG channels (144.9 ± 42.3). Using picture naming (82 patients), BESPoC localized HGM language sites with sensitivity 0.88 ± 0.04, specificity 0.87 ± 0.03, and AUC 0.93 ± 0.02. Using story listening task (55 patients), BESPoC localized HGM language sites with sensitivity 0.64 ± 0.20, specificity 0.89 ± 0.10, and AUC 0.78 ± 0.14. Using the BESPoC algorithm, a group-level map of spontaneous conversation (45 patients) was constructed for expressive and receptive language topography.
Conclusions:
BESPoC was validated against clinical standard-of-care expressive and receptive language tasks, reliably predicting conventional HGM results. Applied to conversation data, BESPoC localized expressive and receptive language similarly to conventional tasks. These results suggest validity of BESPoC for mapping neuronal substrates of language using naturalistic paradigm of spontaneous conversation.
Funding:
NIH NINDS R01 NS115929; CCRF Procter Scholar Award