Novel Speech Signal Processing Algorithms for High-Accuracy Classification of Parkinson's Disease

• Published in: IEEE transactions on biomedical engineering Vol. 59; no. 5; pp. 1264 - 1271
• Main Authors: Tsanas, Athanasios, Little, Max A., McSharry, Patrick E., Spielman, Jennifer, Ramig, Lorraine O.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.05.2012; Institute of Electrical and Electronics Engineers
• Subjects: Aged; Aged, 80 and over; Applied sciences; Biological and medical sciences; Case-Control Studies; Classification algorithms; Decision support tool; Decision Trees; Degenerative and inherited degenerative diseases of the nervous system. Leukodystrophies. Prion diseases; Dysphonia - classification; Dysphonia - physiopathology; Entropy; Exact sciences and technology; feature selection (FS); Female; Frequency measurement; Humans; Information, signal and communications theory; Male; Medical sciences; Mel frequency cepstral coefficient; Middle Aged; Nervous system (semeiology, syndromes); Nervous system as a whole; Neurology; Noise; Noise measurement; Nonlinear Dynamics; nonlinear speech signal processing; Parkinson Disease - classification; Parkinson Disease - diagnosis; Parkinson Disease - physiopathology; Parkinson's disease (PD); random forests (RF); Signal processing; Signal Processing, Computer-Assisted; Speech; Speech processing; Support Vector Machine; support vector machines (SVM); Telecommunications and information theory; Nervous system diseases; Automatic classification; support vector machines (SVM); Decision support system; High precision; Parkinson's disease (PD); feature selection (FS); Parkinson disease; Support vector machine; Signal classification; Decision support tool; Cerebral disorder; random forests (RF); Algorithm performance; Central nervous system disease; Signal processing; Degenerative disease; nonlinear speech signal processing; Non linear processing; Speech processing; Extrapyramidal syndrome; Biomedical engineering
• DOI: 10.1109/TBME.2012.2183367
• ISSN: 0018-9294; 1558-2531

There has been considerable recent research into the connection between Parkinson's disease (PD) and speech impairment. Recently, a wide range of speech signal processing algorithms (dysphonia measures) aiming to predict PD symptom severity using speech signals have been introduced. In this paper, we test how accurately these novel algorithms can be used to discriminate PD subjects from healthy controls. In total, we compute 132 dysphonia measures from sustained vowels. Then, we select four parsimonious subsets of these dysphonia measures using four feature selection algorithms, and map these feature subsets to a binary classification response using two statistical classifiers: random forests and support vector machines. We use an existing database consisting of 263 samples from 43 subjects, and demonstrate that these new dysphonia measures can outperform state-of-the-art results, reaching almost 99% overall classification accuracy using only ten dysphonia features. We find that some of the recently proposed dysphonia measures complement existing algorithms in maximizing the ability of the classifiers to discriminate healthy controls from PD subjects. We see these results as an important step toward noninvasive diagnostic decision support in PD.