OBJECTIVES: The purpose of this study was to investigate the effects of bilateral subthalamic nucleus deep brain stimulation on the phonation of patients with Parkinson's disease in three drug-free conditions: (1) stimulation off, (2) with clinically optimised stimulation parameters, and (3) subthreshold overstimulation, in order to detect differences following voice analysis. PATIENTS AND METHODS: Conversational speech and sustained vowel sounds /a/, /i/, /o/, /u/ and high /i/ were recorded from 22 PD patients. Perceptual analysis, perturbation jitter, shimmer, noise-to-harmonics ratio, and nonlinear dynamic analysis (NDA) with detrended fluctuation analysis and recurrence period density entropy were measured and compared to the above conditions. Quadratic discriminant analysis (QDA) was used to investigate stimulation conditions for given acoustic data. RESULTS: The changes of perturbation measurements for the above conditions were not significant. With differences between vowels, NDA showed more significant changes and more powerful correlation with perceptual scores than perturbation measurements. NDA was significantly more sensitive during the QDA of the conditions. CONCLUSIONS: Acoustic voice analysis of sustained vowels can help with recognizing the overstimulated condition, and, with an appropriate test battery and software package including nonlinear dynamic analysis, it can be a valuable tool for fine adjustments of stimulation parameters.
OBJECTIVES: The purpose of this study was to investigate the effects of bilateral subthalamic nucleus deep brain stimulation on the phonation of patients with Parkinson's disease in three drug-free conditions: (1) stimulation off, (2) with clinically optimised stimulation parameters, and (3) subthreshold overstimulation, in order to detect differences following voice analysis. PATIENTS AND METHODS: Conversational speech and sustained vowel sounds /a/, /i/, /o/, /u/ and high /i/ were recorded from 22 PDpatients. Perceptual analysis, perturbation jitter, shimmer, noise-to-harmonics ratio, and nonlinear dynamic analysis (NDA) with detrended fluctuation analysis and recurrence period density entropy were measured and compared to the above conditions. Quadratic discriminant analysis (QDA) was used to investigate stimulation conditions for given acoustic data. RESULTS: The changes of perturbation measurements for the above conditions were not significant. With differences between vowels, NDA showed more significant changes and more powerful correlation with perceptual scores than perturbation measurements. NDA was significantly more sensitive during the QDA of the conditions. CONCLUSIONS: Acoustic voice analysis of sustained vowels can help with recognizing the overstimulated condition, and, with an appropriate test battery and software package including nonlinear dynamic analysis, it can be a valuable tool for fine adjustments of stimulation parameters.