Electroencephalography and local field potentials in Parkinson's disease: models and diagnosis

Parkinson’s disease (PD) is a multimodal neurodegenerative disorder caused by the death of the dopamine-containing neurons resulting in impaired movement and cognition. This loss of midbrain dopamine neurons that project throughout the brain, including the cerebral cortex and Basal Ganglia causes profound changes in cortical and subcortical brain activity as measured by Electroencephalography (EEG) or intracranial recordings of local field potentials (LFP). This doctoral work focuses on these changes in brain activities captured by EEG and LFP data to develop a classification method for PD detection and propose a computational model for the neuronal networks in the Basal Ganglia region for a better understanding of these changes.

In particular, this doctoral work proposes a generation of novel features based on Linear predictive coding (LPC) which can efficiently detect PD-related changes in EEG and LFP recordings. Based on these LPC-based features, classification methods are proposed for diagnosing PD with EEG, detecting the effects of dopamine modulation in rodent models of PD, capturing cortical changes due to Deep brain stimulation (DBS) and achieving a classification index with high correlations with PD-related non-motor symptomatic score. This doctoral work also presents an axiomatic approach to Jensen-Rit neural mass models and formulates generalized theoretical estimations of the output power spectral density. Furthermore, models for two feedback networks of Basal ganglia are proposed and validated using LFP recordings from humans and rodent models of PD. Finally, the changes in the model due to PD are discussed and a theoretical understanding of dominant peak frequency is provided.