Abstracts

Rationale: Brain processes underlying verbal memory encoding can be probed in the spectral activities of intracranial EEG (iEEG) recordings during the presentation of words in a free recall task. These spectral activities can be detected on active electrodes classified using machine learning. Previous studies used spectral power to analyze spatiotemporal patterns observed on the active electrodes from patients admitted to the Epilepsy Monitoring Unit (EMU). However, the stress and distractions of the hospital environment may affect patient performance on cognitive tasks. The purpose of this study is to develop a platform for the investigation of the spectral, spatial, and temporal properties of brain activities underlying human verbal memory encoding in ambulatory, human subjects outside the hospital. Methods: Intracranial recordings were obtained during the encoding of words from implanted subdural and depth electrodes in two groups of patients: those undergoing surgical evaluation of drug-resistant epilepsy in the EMU and those implanted with a permanent RC+S device for ambulatory EEG monitoring with electrodes in the hippocampus and anterior nucleus of the thalamus. Participants were instructed to study lists of words presented on a tablet screen for a delayed free recall test. Lists of 12 nouns were chosen at random and without replacement from a pool of common English. After a distractor task, participants were asked to recall as many of the words as possible. Each word presentation epoch iEEG was filtered (1000-order Barlett-Hanning) before spectral decomposition and normalization. Results: For each electrode, we determined the average spectral power at each time point across all word presentation epochs and found the overall induced power during word presentation. Electrodes were classified into active and inactive clusters. We analyzed the effects of frequency band, brain region (Brodmann Area), and test performance on the overall induced power during the verbal memory task. Our analysis of the spectral, spatial, and temporal patterns recorded on active electrodes during verbal memory encoding revealed consistent and robust differences in low theta (2-4 Hz), high theta (5-9 Hz), alpha (10-15 Hz), beta (16-25 Hz), low gamma (25-55 Hz), and high gamma (65-115 Hz) frequency bands. Brodmann areas within the occipital lobe were the most likely to be active in patients within the EMU. We also observed distinct temporal patterns of the induced power change characteristic to the low and high frequency bands studied. A clear subsequent memory effect between recalled and forgotten trials has been found from tasks in patients in the EMU. The first RC+S subject has been recruited, and we have used our experiences in the EMU to successfully set up a tablet-based platform for implementation with the ambulatory subjects. Conclusions: Analysis of human iEEG reveals that human verbal memory encoding can be characterized by the spectral, spatial, and temporal properties of brain activity. Verbal memory work with EMU patients has allowed us to develop a mobile platform for cognitive assessment of chronically implanted patients outside the hospital. These analyses increase our understanding of verbal memory processing in the brain and will help us understand the effects of living with epilepsy, recurrent seizures, and medications. Funding: DARPA RAM Project N66001-14-2-4032NIH UH2/UH3-NS95495First Team Programme of the Foundation for Polish Science co-financed by the European Union under the European Regional Development Fund