Speech scan for Parkinson's

Parkinson’s disease “could be diagnosed by voice changes”, according to The Daily Telegraph. The newspaper said that Parkinson’s disease could be diagnosed earlier by testing for subtle changes in speech that often accompany the condition.

This news story was based on research that compared different ways of analysing the sound wave patterns generated when vowels are spoken. The researchers found that one method could detect changes in articulation that were present in people with Parkinson’s disease but not a comparison group of healthy individuals.

One point to note is that the participants with Parkinson’s disease in the study had been diagnosed approximately seven years before this research, so their disease may have been fairly advanced. While this work encourages further research into this area, it remains to be seen to see whether the technique is sensitive enough to detect changes in articulation that may occur very early in the disease. Further research will be needed to judge whether this technique will lead to earlier diagnoses of Parkinson’s disease.

Where did the story come from?

This research was carried out by Dr Shimon Sapir from the University of Haifa, Isreal, and colleagues at the National Center for Voice and Speech in Denver, Colorado, in the US. The study was funded by the National Institute on Deafness and Other Communication Disorders in the US. The study was published in a peer-reviewed medical journal, the Journal of Speech, Language and Hearing Research.

The Daily Telegraph focussed on the potential of voice articulation analysis to diagnose Parkinson’s disease. The patients who participated in the study had, on average, been diagnosed approximately seven years before. Further research would be required to assess whether this technique could be used to detect changes in articulation earlier in the disease.

What kind of research was this?

People with Parkinson’s disease can develop a type of speech disorder called dysarthria. This happens when the condition affects the parts of the brain that control the movements needed for speech. Dysarthria is characterised by poor articulation. It has been suggested that if measuring the severity of dysarthria is possible, it could be used to monitor deterioration or improvement of the condition due to disease progression or treatments for Parkinson’s disease.

This non-randomised controlled study tested the ability of an acoustic analysis technique, called the Formant Centralization Ratio (FCR), to measure how much speech was affected in people with Parkinson’s disease. The researchers wanted to assess whether FCR was better than an existing technique, called the Vowel Space Area (VSA) method, to distinguish dysarthric speech from healthy speech.

When we articulate vowels, two sound wave frequency patterns (formants) are generated. These sound wave patterns change in a predictable way when the mouth and tongue are moved to make the different sounds that combine to form vowels. Most types of dysarthria are characterised by a reduced range of articulatory movement and the resulting changes in the frequency of the formants compared to normal speech. The FCR and VSA analysis methods use different mathematical models to analyse sound wave patterns in speech.

What did the research involve?

This small study included 38 individuals with Parkinson’s disease. Nineteen of these individuals had received intensive voice/speech therapy (the treatment group) and the other 19 had received no treatment (the non-treatment group). These groups were compared to 14 healthy individuals (control subjects), who were matched for age and sex. All participants spoke American English as their first language and were recruited mostly from Tuscon in Arizona or Denver in Colorado.

Most of the patients with Parkinson’s disease had dysarthria that was rated as moderate or mild, and was characterised by hoarseness, monotone speech and reduced loudness. The average number of years since diagnosis of Parkinson’s was approximately seven years.

The participants in the treatment group were tested before they received speech therapy and again after the therapy. The non-treatment group that did not receive speech therapy and the healthy control group were tested on the same days as the treatment group.

The participants were asked to repeat phrases, such as: “The blue spot is on the key”, “The potato stew is in the pot” and “Buy Bobby a puppy”. Their voices were recorded using a microphone positioned 6cm from their lips, which was either directly linked to a computer or to a digital recorder linked to a computer. Vowels were extracted from certain words, including “key”, “stew”, “Bobby”, and “pot”, and the sound wave patterns were analysed.

What were the basic results?

In the first set of recordings (before speech training), FCR analysis could detect a difference between the healthy control group and the two Parkinson’s groups (treatment and non-treatment groups). It detected no difference between the two Parkinson’s disease groups. The VSA analysis method did not detect a difference between the groups.

VSA could detect differences between men and women’s speech, whereas FCR could not.

Both VSA and FCR could detect differences in voices of the treatment group after their treatment, but the FCR analysis was more robust in detecting these differences.

How did the researchers interpret the results?

The researchers suggest that, although their research should be considered as preliminary, FCR is a valid and highly sensitive method of measuring normal and abnormal vowel articulation. They also say that its performance is superior to that of VSA in distinguishing dysarthric speech from healthy speech.

Conclusion

This preliminary research has demonstrated that the FCR analysis method can be used to detect dysarthric speech in people with Parkinson’s disease, and may be superior to the VSA analysis method sometimes currently used. However, the researchers say that there are other techniques to assess dysarthria which they did not test against the FCR method. Therefore, they cannot say that FCR is the overall preferred tool for assessing dysarthria without further research.

As the participants for this study were not randomised to their groups (treatment, non-treatment and healthy control groups), it is possible that the people selected differed in important ways that were not due to either treatment or disease. The researchers provide very little detail about the characteristics of the people selected, who appear to have been of a similar age and stage of disease (or time from diagnosis) in the parkinson’s disease groups. This means that it is not possible to say whether other factors, such local dialects, could explain the differences in this study.

The study also looked at a relatively small population with Parkinson’s disease. Other conditions, such as motor neurone disease or cerebral palsy, may result in dysarthria. The researchers say that further research would be necessary to test how well FCR could assess dysarthria secondary to these conditions because the type of movement impairment involved may differ from that seen in Parkinson’s disease.

The FCR method detected a difference in individuals’ vowel sound wave patterns after speech therapy. It has been suggested that analysing articulation with techniques such as FCR could be used to monitor disease progression or response to treatment. This study warrants further research into whether FCR is sensitive enough to detect changes in articulation over time and how soon after the onset of Parkinson’s disease these changes can be detected. The results of such further studies may indicate whether the technique could be used as a diagnostic tool in future.