Classification of Parkinson's disease Using Pitch Synchronous Speech Analysis

Area of Science:

• Computational linguistics
• Neuroscience
• Biomedical engineering

Background:

• Parkinson's disease (PD) significantly impacts speech production, affecting cognitive and muscular functions.
• Computational analysis of speech is a key research area for identifying PD-related speech perturbations.
• Current methods focus on extracting speech features to detect these changes.

Purpose of the Study:

• To propose and evaluate a novel pitch synchronous methodology for speech feature extraction and analysis in Parkinson's disease.
• To address the dimensionality challenge in feature extraction through feature selection.
• To utilize unsupervised k-means clustering for classification of PD patients.

Main Methods:

• Features are extracted from individual speech pitch cycles.
• Analysis is performed on the variances of these features, creating a pitch synchronous approach.
• Feature selection is employed to reduce dimensionality.
• Unsupervised k-means clustering is used for classification.

Main Results:

• The proposed pitch synchronous methodology demonstrated promising results in analyzing speech data.
• The approach effectively captured speech perturbations associated with Parkinson's disease.
• Classification using k-means clustering showed the potential of this method.

Conclusions:

• Pitch synchronous speech analysis is a viable and effective approach for Parkinson's disease detection.
• This novel methodology offers a promising tool for objective assessment of PD.
• Further research can build upon this technique for improved diagnostic capabilities.