ETD

• Electroencephalography (EEG)
• Parkinson's Disease
• Speech impairment

Speech impairment is a common feature of Parkinson’s disease (PD), most commonly manifesting as hypokinetic dysarthria, yet its cortical mechanisms remain incompletely understood. This thesis aimed to identify non-invasive electroencephalographic (EEG) markers of speech dysfunction. EEG was recorded from 7 PD patients during resting state, reading and monologue tasks. Spectral analysis, assessing relative and relative periodic power via cluster-based permutation testing at sensor and source levels, was performed. Additionally, functional connectivity was evaluated using edge-level imaginary coherence and global network metrics. These neural features were subsequently correlated with extracted speech metrics.
Spectral analysis revealed that the transition from rest to overt speech is associated with an increase in delta and gamma power and a widespread desynchronization in the alpha and beta bands. Spectral parametrization isolated periodic activity, revealing that scripted reading induces a significantly greater suppression of theta and alpha periodic power compared to spontaneous speech. Furthermore, functional connectivity analysis demonstrated that reading reduced global network integration.
Notably, the magnitude of alpha and low beta spectral modulation during speech was positively associated with articulatory precision and speech intelligibility measures, whereas baseline periodic theta and alpha power measured at rest emerged as predictors of speech performance across both tasks. Moreover, greater network integration at rest was related to higher intelligibility of speech.
Together, these findings highlight cortical spectral and topological dynamics as promising non-invasive correlates of speech impairment in PD.