Consciousness, Cognition, and Connectivity: An Investigation in Focal Epilepsy

Abstract

Epilepsy affects approximately 50 million people worldwide, often resulting in recurrent consciousness-impairing seizures and chronic neurocognitive deficits. Despite advancements in medical therapy, at least 40% of patients continue to experience significant impairments in daily functioning. While neurosurgical interventions can offer seizure freedom and neurocognitive improvement, successful outcomes are limited to 60-70% of patients, highlighting the need for more effective treatments. This dissertation investigates brain networks in epilepsy to advance understanding mechanisms of consciousness-impairing seizures and neurocognitive deficits. The research builds upon two key hypotheses: the Network Inhibition Hypothesis, which posits that focal seizures disrupt subcortical structures critical for cortical activation, resulting in impaired consciousness, and the Extended Network Inhibition Hypothesis, which suggests that recurrent seizure propagation to these subcortical arousal structures leads to chronically reduced cortical activation and subsequent neurocognitive deficits. These detrimental sequalae of epilepsy can be studied in humans during seizures with intracranial recordings and between seizures with functional magnetic resonance imaging (fMRI). First, as a pre-requisite to study subcortical structures important in consciousness we developed a novel method using MRI and deep learning to capture patient-specific subcortical anatomy. Second, to augment our understanding of consciousness we then analyzed intracranial recordings from consciousness-impairing and consciousness-sparing seizures to identify network signatures of impaired consciousness. Third, we sought to improve our understanding of the effect of recurrent consciousness-impairing seizures on arousal networks and neurocognitive deficits with fMRI. Fourth, we hoped to improve seizure and neurocognitive outcomes with a machine learning approach for localization of seizure onset zones. Fifth, to improve our understanding of network signatures of neurocognitive impairments we used data-driven methods to uncover brain states linked to neurocognitive impairment. By combining both data-driven and hypothesis-driven methods with intracranial recordings and fMRI we sought to gain further knowledge on the underlying pathology behind consciousness-impairment and neurocognitive deficits in epilepsy. We believe that this work lays the foundation for the development of novel approaches to decrease the morbidity of epilepsy.