Advisor
Mark A. Clements, ScD (Georgia Institute of Technology)

Thesis Committee
Thomas Wichmann, MD (Emory University)
Elizabeth Buffalo, PhD (Emory University)
Chris Rozell, PhD (Georgia Institute of Technology)
Garrett Stanley, PhD (Georgia Institute of Technology)

Parkinson's Disease (PD) is a neurodegenerative disorder that affects approximately 1 million people in the United States. The disease is characterized by progressive motor and non-motor disability and loss of dopaminergic cells in the substantia nigra. There is no cure for PD, although carefully managed pharmacological treatments can typically control symptoms for 5 to 10 years. Unfortunately, these medications become less effective over time, and are often accompanied by adverse effects.  Better monitoring of the severity of parkinsonism, and a clearer understanding of the pathophysiology of parkinsonism and the effects of treatments may significantly improve the quality of life for patients.

This proposed work investigates electrophysiologic and movement abnormalities in rhesus monkeys and humans. Novel signal processing applications of cross frequency coupling, feature selection, and canonical correlation are developed to discover the most significant electrophysiological changes in parkinsonism and to allow technology-assisted multi-modal grading of the disease. The signal processing algorithms are currently being implemented in off-the-shelf hardware to develop a new system for remote monitoring of PD.

Four preliminary findings are described in the proposal, based on results obtained in monkeys rendered parkinsonian with the dopaminergic neurotoxin MPTP. First, feature selection and classification methods are useful for identifying the most important electrophysiological changes in parkinsonism as compared to the baseline condition. Second, in parkinsonian monkeys, beta oscillations have an increased influence on the amplitude of oscillations in other frequencies, even in the absence of elevated average power in the beta band.  Third, certain features from primary motor cortex (M1) electroencephalograms (EEGs) and subthalamic nucleus (STN) local field potentials (LFPs) correlate with each other and with the severity of parkinsonism motor scores. Fourth, phase-amplitude cross-frequency coupling (CFC) measures taken from M1 EEGs and STN LFPs are useful for assessing degree of parkinsonism.