Advances in Parkinson's Disease detection and assessment using voice and speech: A review of the articulatory and phonatory aspects

https://doi.org/10.1016/j.bspc.2021.102418Get rights and content

Abstract

Parkinson's Disease (PD) affects speech in the form of dysphonia and hypokinetic dysarthria. Multiple studies have evaluated PD's influence on different aspects of speech, showing differences between speakers with and without PD. Most recent studies are focused on the proposal of new automatic and objective tools to help in the diagnosis and severity assessment. This comprehensive review identifies the most common features and machine learning techniques employed in automatically detecting and assessing the severity of PD using phonatory and articulatory aspects of speech and voice. We discuss their discriminant properties and literature findings as well as identify common methodological issues that can potentially bias results. The objective is to provide a broad overview of these methods, their advantages and disadvantages, and to identify the most promising methodologies to be explored in future works. We conclude that there is clear evidence that the articulatory and phonatory aspects of speech and voice are relevant for the automatic detection and severity assessment of PD. However, there is no standard methodology sufficiently validated in a clinical trial, and further research is required, especially to develop larger corpora and identify new objective biomarkers.

Introduction

Parkinson's Disease (PD) is a chronic and progressive condition caused by the gradual neuronal death in the substantia nigra, implicated in the production of dopamine neurotransmitters, which play a crucial role in motor control. [1]. Despite the fact that PD is the second most common neurodegenerative disease, the average diaganosis time is above two years [2]. Therefore, new precision medicine tools based on patient's signs are needed to assist diagnosis and personalized treatment. To this respect, speech involves complex and precise coordination of the respiratory system, larynx, and supraglottal articulators, being an excellent candidate to provide such diagnostic information [3].

Several studies previous to 1970 described the speech of people with PD [4], [5], although it is not until 1969 [6], [7] that a group of researchers analyzed more thoroughly the problems of phonation, prosody, and articulation of PD patients. Since then, a plethora of studies have presented evidence that the neurodegenerative processes associated with PD cause dysphonia and dysarthria, particularly hypokinetic dysarthria [6], [8], [9], [10] in different stages of the disease [11]. Dysphonia can be defined as the speaker's incapacity to produce a normal phonation due to the phonatory system's impaired functioning, while dysarthria is more related to problems with articulation when pronouncing words. More specifically, hypokinetic dysarthria is characterized by a reduction of loudness and articulation amplitude, slow speech rates combined with rushes of fast speech sometimes, and a decrease of intelligibility mostly. Some specialists suggest that 90% of PD patients suffer from dysarthria [12], [3] after a median latency period of 7 years since diagnosis, which is in accordance with the study [13], although other works such as [14], [15] point to lower values. The prevalence of dysarhtria and dysphonia in PD patients is unclear. Some studies determine a higher prevalence of dysphonia (65.5%) over articulatory impairment (38.5%) when perceptually evaluating 200 patients [15]. These results are consistent with those reported by previous studies [13] employing the same number of patients. However, the mentioned studies employed perceptual methods. More recent studies such as [16] suggest a higher prevalence of articulatory deficits. Whereas this last study uses objective measurements, the employed cohort is much smaller. In this context, literature shows a myriad of studies and approaches investigating the influence of PD in the voice or speech of patients and proposing new biomarkers or automatic detection schemes to support the diagnosis of this condition. The interest in parkinsonian speech has increased with time as shown in Fig. 1, which includes the number of studies per year related to dysarthria in PD and the influence of PD in motor speech aspects between 1956 and May 2020, according to PubMed. In general terms, it is possible to divide these studies into four main groups depending on the analyzed speech aspect, i.e., phonatory, articulatory, prosodic, and cognitive-linguistic:

  • Phonatory studies are related to the glottal source and resonant structures of the vocal tract, employing most of the time sustained vowels as acoustic material for the analysis.

  • Studies based on articulatory aspects are more diverse as there exist more analysis possibilities: the features or the acoustic measurements analyzed can be extracted from different types of sound segments and can be related to the velocity or acceleration of articulators, type of transitions between segments, or the evolution of formants among others.

  • Prosodic studies mainly focus on paralinguistic features such as pitch variation, syllable rate analysis, or the manifestation of emotions in the speech signal.

  • Finally, the cognitive-linguistic approaches analyze the deviations in cognitive behavior by examining the vocabulary, sentence complexity, phrase construction, and the existence of word repetitions, among other manifestations.

The speech task or acoustic material that is used in each case is a differentiating key factor of the four aspect groups. In the phonatory analysis, sustained vowels are commonly employed as acoustic material, while in the other three groups, the use of connected speech is necessary. For the latter, monologues, read passages, and Diadochokinetic (DDK) tasks1 are mostly considered.

Among all the studies evaluating of the patient's speech, it is possible to differentiate between two types of analysis:

  • Perceptual assessment, where trained evaluators follow a specific protocol such as the Frenchay Dysarthria assessment [17] to rate certain aspects of the speech

  • Objective analysis, which makes use of algorithms to characterize the signal through certain acoustic features.

Some of the works performing objective assessment propose measures or biomarkers to assist PD diagnosis or correlate with its severity. Others propose automatic binary classifiers to distinguish between speakers with and without PD or automatically predict the disease's stage by processing the speech signal.

To this extent, the literature about PD presents reviews concerning the influence of speech treatments [18], [19], the effects of the pharmacological and surgical therapy [20], [21], [22] on voice and speech, prosody [23], aspects of language [24], speech perception [25], or behavioral treatments for speech [26], among others. Despite that the studies in the literature about parkinsonian speech are abundant, as Fig. 1 shows, no article reviews the findings on the use of the phonatory and articulatory aspects of voice and speech with diagnosis purposes.

Consequently, this article aims to fill this gap, presenting a comprehensive review of the state of the art on PD diagnosis and severity assessment employing signal processing and machine learning techniques focused on phonatory and articulatory aspects of voice and speech. Our objective is to perform a survey of the state of the art to provide a categorization of those studies describing the main speech tasks, features, and signal processing techniques employed in the objective assessment of PD. This review aims to provide a broad overview and discussion of the existing methods and their advantages/disadvantages, and will help to identify the most recommended methodologies to be followed in future works. As secondary objectives, we identify some methodological issues that often arise in the processing of PD speech, and describe the corpora that are publicly available to analyze phonatory and articulatory aspects of parkinsonian speech.

The article is organized as follows: Section 2 introduces the methods to the review, Sections 3 and 4 includes an analysis of phonatory and articulatory aspects, respectively, while Section 5 lists and describes the most common features employed in the state of the art. Lastly, Sections 6 and 7 include the discussion and conclusions of the article.

Section snippets

Methods

In this review, we selected articles from PubMed and Google Scholar search after having considered the following selection criteria: documents in English analyzing the effect of PD in voice and speech, proposing new biomarkers or methodologies to detect or assess the disease utilizing speech technologies. The search queries employed to obtain the list of studies can be found in the additional material accompanying this paper. As the review is focused on the articulatory and phonatory aspects of

Evidence of the influence of PD in the phonatory system

Studies about the influence of PD in the phonatory system mainly analyze the impairments in phonatory-related structures and muscles like the diaphragm, the muscles connected to the larynx, the vocal folds, or the supra-glottal resonant cavities. The type of acoustic material analyzed in these cases is usually the voice signal from one or several sustained vowels, which can be maintained during a specific time-lapse or modulated in frequency and amplitude. Sustained vowels are expected to

Evidence of PD's influence in phoneme articulation

Connected speech contains fluctuations in vocal characteristics such as voice onset, voice termination, and voice breaks, which are considered crucial in quality of voice and speech evaluation. These characteristics are not fully represented with sustained vowels. Several studies in literature point out to an evident influence of PD in phoneme articulation, and study that influence in some specific sounds or articulatory movements. In one of the earliest studies trying to determine parkinsonian

Common features employed for detection or severity assessment of PD

This section lists some of the most common features employed in the automatic detection and severity assessment of PD. Table 1 includes a list of some of the motor feature families and coefficients analyzed in the literature to detect or assess PD. The table includes comments about the significance of these features’ discriminative properties and a reference to the studies using them. All the features in the list have been presented in this article, and the most relevant are discussed in the

Discussion

In this review, we have analyzed different studies dealing with the influence of PD on the phonatory and articulatory aspects of voice and speech and how such influence can help to design new tools to assist the diagnosis and evaluation of the disease. The literature reports a large amount of features and methods to characterize parkinsonian speech. Some studies employ feature vectors with hundreds of measurements that are used as the input of machine learning classifiers to create automatic

Conclusions

In this study, we have presented a comprehensive review of state of the art on the use of speech analyzing the phonatory and articulatory aspects to support PD diagnosis and evaluation, with a particular emphasis on those studies proposing approaches to detect PD automatically or to assess its severity. In view of the state of the art, and from an analysis of better and worse practices, we also provide recommendations to be followed in the future.

The aforementioned works are the pillars that

Authors’ contributions

Laureano Moro-Velazquez: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Validation; Visualization; Roles/Writing – original draft; Writing – review & editing.

Jorge A. Gomez-Garcia: Data curation; Resources; Software; Validation; Visualization; Roles/Writing – original draft; Writing – review & editing.

Julian D. Arias-Londoño: Formal analysis; Investigation; Validation; Writing – review & editing.

Najim Dehak: Formal analysis; Funding acquisition; Project

Acknowledgements

This work was supported by the Ministry of Economy and Competitiveness of Spain under grant DPI2017-83405-R1.

Declaration of Competing Interest

The authors report no declarations of interest.

References (208)

  • A. Bayestehtashk et al.

    Fully automated assessment of the severity of parkinson’s disease from speech

    Comput. Speech. Lang.

    (2015)
  • C.O. Sakar et al.

    A comparative analysis of speech signal processing algorithms for parkinson’s disease classification and the use of the tunable q-factor wavelet transform

    Appl. Soft Comput.

    (2019)
  • L. Moro-Velazquez et al.

    Analysis of speaker recognition methodologies and the influence of kinetic changes to automatically detect parkinson’s disease

    Appl. Soft Comput.

    (2018)
  • L. Moro-Velazquez et al.

    A forced gaussians based methodology for the differential evaluation of parkinson’s disease by means of speech processing

    Biomed. Signal Process. Control.

    (2019)
  • M.P. Caligiuri

    The influence of speaking rate on articulatory hypokinesia in parkinsonian dysarthria

    Brain Lang.

    (1989)
  • R.F. Pfeiffer et al.

    Parkinson's Disease

    (2013)
  • A.J. Hughes et al.

    The accuracy of diagnosis of parkinsonian syndromes in a specialist movement disorder service

    Brain

    (2002)
  • J.R. Duffy

    Motor Speech Disorders: Substrates, Differential Diagnosis, and Management

    (2013)
  • H.A. Kaplan et al.

    A study of the effectiveness of drug therapy in parkinsonism

    J. Nerv. Ment. Dis

    (1954)
  • M. Sarno

    Speech impairment in parkinson’s disease

    Arch. Phys. Med. Rehabil.

    (1968)
  • F.L. Darley et al.

    Differential diagnostic patterns of dysarthria

    J. Speech Lang. Hear. Res.

    (1969)
  • F.L. Darley et al.

    Clusters of deviant speech dimensions in the dysarthrias

    J. Speech Lang. Hear. Res.

    (1969)
  • H. Ackermann et al.

    Articulatory deficits in parkinsonian dysarthria: an acoustic analysis

    J. Neurol. Neurosurg. Psychiatry

    (1991)
  • P. Blanchet et al.

    Speech rate deficits in individuals with parkinson’s disease: a review of the literature

    J. Med. Speech Lang. Pathol.

    (2009)
  • S. Polychronis et al.

    Speech difficulties in early de novo patients with parkinson’s disease

    Parkinsonism Relat. Disord.

    (2019)
  • J. Müller et al.

    Progression of dysarthria and dysphagia in postmortem-confirmed parkinsonian disorders

    Arch. Neurol.

    (2001)
  • J.A. Logemann et al.

    Frequency and cooccurrence of vocal tract dysfunctions in the speech of a large sample of parkinson patients

    J. Speech Hear. Disord.

    (1978)
  • L. Hartelius et al.

    Speech and swallowing symptoms associated with parkinson’s disease and multiple sclerosis: a survey

    Folia Phoniatr. Logop.

    (1994)
  • A.K. Ho et al.

    Speech impairment in a large sample of patients with parkinson’s disease

    Behav. Neurol.

    (1999)
  • J. Rusz et al.

    Speech disorders reflect differing pathophysiology in parkinson’s disease, progressive supranuclear palsy and multiple system atrophy

    J. Neurol.

    (2015)
  • P. Enderby

    Frenchay dysarthria assessment

    Brit. J. Disord. Commun.

    (1980)
  • L.O. Ramig et al.

    Speech treatment for parkinson’s disease

    Expert Rev. Neurother.

    (2008)
  • M. Trail et al.

    Speech treatment for parkinson's disease

    NeuroRehabilitation

    (2005)
  • S. Skodda

    Effect of deep brain stimulation on speech performance in parkinson’s disease

    Parkinson’s Dis.

    (2012)
  • L. Brabenec et al.

    Speech disorders in parkinson’s disease: early diagnostics and effects of medication and brain stimulation

    J. Neural Transm.

    (2017)
  • S. Skodda et al.

    Speech rate and rhythm in parkinson’s disease

    Mov. Disord.

    (2008)
  • D. Ingvar

    Serial aspects of language and speech related to prefrontal cortical activity. a selective review

    Hum. Neurobiol.

    (1983)
  • L.C. Kwan et al.

    Perception of speech by individuals with parkinson’s disease: a review

    Parkinson’s Dis.

    (2011)
  • C. Atkinson-Clement et al.

    Behavioral treatments for speech in parkinson’s disease: meta-analyses and review of the literature

    Neurodegener. Dis. Manag.

    (2015)
  • L. Moro-Velazquez et al.

    Study of the automatic detection of parkison's disease based on speaker recognition technologies and allophonic distillation

  • K. Elemetrics

    Multi-Dimensional Voice Program (MDVP).[Computer program.]

    (1993)
  • P. Boersma

    Praat: Doing Phonetics By Computer

    (2006)
  • F. Eyben et al.

    Opensmile: the munich versatile and fast open-source audio feature extractor

  • D.G. Hanson et al.

    Cinegraphic observations of laryngeal function in parkinson’s disease

    Laryngoscope

    (1984)
  • I. Midi et al.

    Voice abnormalities and their relation with motor dysfunction in parkinson’s disease

    Acta Neurol. Scand.

    (2008)
  • J. Mekyska et al.

    Assessing progress of parkinson's disease using acoustic analysis of phonation

  • M.J. Hammer et al.

    Laryngeal somatosensory deficits in parkinsonś disease: implications for speech respiratory and phonatory control

    Exp. Brain Res.

    (2010)
  • P.H. Dejonckere et al.

    A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques

    Eur. Arch. Oto-Rhino-L

    (2001)
  • L.A. Ramig et al.

    Acoustic analysis of voices of patients with neurologic disease: rationale and preliminary data

    Ann. Otol. Rhinol. Laryngol.

    (1988)
  • J. Rusz et al.

    Quantitative acoustic measurements for characterization of speech and voice disorders in early untreated parkinson’s disease

    J. Acoust. Soc. Am.

    (2011)
  • Cited by (79)

    View all citing articles on Scopus
    View full text