One of the high priority research areas listed in the BRAIN 2025 Report is a better understanding of the brain network dynamics across time and space from human electrophysiological recordings. The RFA-MH-20- 120 ?BRAIN Initiative: Secondary Analysis and Archiving of BRAIN Initiative Data? specifically focuses on the use of large volumes of existing data stored in open access public databases. In the proposed studies, we will perform an innovative secondary analysis of existing intracranial EEG records from patients with medicine- resistant epilepsy collected during their presurgical evaluation. These records from public repositories would allow us to study dynamic interactions between hippocampal and neocortical structures during the development of pathological activity as well as during verbal memory tasks. We will use a novel and rapidly developing approach from the realm of machine learning algorithms namely, deep learning neural networks. Physiological mechanisms of memory formation and consolidation as well as pathophysiological mechanisms of epilepsy are poorly understood. The main hypothesis of this proposal is that physiological function and dysfunction of the hippocampal-neocortical system may have a common mechanism that depends on the dynamic interaction of electrophysiological oscillations in the system. Brain oscillations have been suggested to be involved in information transfer within and between brain networks by modulating neural excitability at different spatial and temporal scales. The interaction of oscillations across different time scales is referred to as ?cross-frequency coupling? and it represents a high-order structure in the functional organization of brain rhythms. Aberrant patterns of cross-frequency coupling may lead to memory dysfunction as well as facilitate the propagation of pathological activity. At present, the field does not have a mechanistically justified criterion to distinguish between physiological and pathological oscillations especially regarding their high-order interactions. We will study cross- frequency coupling during the interictal-ictal transition (Aim 1) and during successful and unsuccessful verbal memory performance (Aim 2) using Deep Learning algorithms. One of the study goals is to create a neural network capable of recognizing patterns of cross-frequency coupling as biomarkers of different physiological and pathological functional states of the brain. A comprehensive characterization of cross-frequency coupling would allow us to distinguish between physiological and aberrant forms of interaction between brain networks at different time scales. The study results will deepen our knowledge about the functional organization of brain rhythms and will provide a new method to recognize functional states from electrophysiological records of brain activity. This method will be applicable for a more accurate monitoring, diagnosis and treatment of memory impairments in neurological and mental disorders as well as for the development of new tools for seizure prediction and control.
A better understanding of brain dynamics from human electrophysiological recordings will help us find new ways to treat neurological and mental disorders and the BRAIN Initiative encourages the secondary analysis of large amounts of existing data. The proposed study will analyze intracranial EEG records from human participants stored in public data repositories using artificial intelligence methods. The results will deepen our knowledge of brain function and dysfunction and will pave the way for the development of new methods to diagnose and treat brain disorders.
