Detail publikace

Automatic voice analysis for dysphagia detection

Originální název

Automatic voice analysis for dysphagia detection

Anglický název

Automatic voice analysis for dysphagia detection

Jazyk

en

Originální abstrakt

Purpose: A videofluoroscopic analysis is the gold standard approach to determine whether a dysphagia problem exists. This procedure is invasive as it involves radiation but also provides the most direct physical evidence of swallowing problems. The main goal of this study was to evaluate an automatic tool based on voice analysis to support medical detection of dysphagia. Methods: An automatic voice analysis system has been developed. Prior to (basal) and immediately following (viscosity) swallowing liquids of varying viscosity and volume, individuals with Parkinson Disease were required to produce the same test word. The acoustic features (linear and non-linear) of this word were then analyzed with regard to specific situations by standard and Machine Learning methods. Results: The results indicated a high degree of accuracy in detecting voice associated with basal and viscosity states. Conclusion: Thus, while the gold standard of dysphagia diagnosis continues to involve video-fluoroscopy analysis, the consideration of voice analysis may prove to be a far simpler and less invasive approach to diagnosis by advanced voice features.

Anglický abstrakt

Purpose: A videofluoroscopic analysis is the gold standard approach to determine whether a dysphagia problem exists. This procedure is invasive as it involves radiation but also provides the most direct physical evidence of swallowing problems. The main goal of this study was to evaluate an automatic tool based on voice analysis to support medical detection of dysphagia. Methods: An automatic voice analysis system has been developed. Prior to (basal) and immediately following (viscosity) swallowing liquids of varying viscosity and volume, individuals with Parkinson Disease were required to produce the same test word. The acoustic features (linear and non-linear) of this word were then analyzed with regard to specific situations by standard and Machine Learning methods. Results: The results indicated a high degree of accuracy in detecting voice associated with basal and viscosity states. Conclusion: Thus, while the gold standard of dysphagia diagnosis continues to involve video-fluoroscopy analysis, the consideration of voice analysis may prove to be a far simpler and less invasive approach to diagnosis by advanced voice features.

BibTex


@article{BUT140606,
  author="Karmele {Lopez-de-Ipina} and Pilar {Calvo} and Marcos {Faúndez Zanuy} and Pere {Clavé} and Weslania Viviane {Nascimento} and Unai {Martinez de Lizarduy} and Alvarez-Berdugo {Daniel} and Arreola García {Viridiana} and Omar {Ortega} and Jiří {Mekyska} and M. Pilar {Sanz-Cartagena}",
  title="Automatic voice analysis for dysphagia detection",
  annote="Purpose: A videofluoroscopic analysis is the gold standard approach to determine whether a dysphagia problem exists. This procedure is invasive as it involves radiation but also provides the most direct physical evidence of swallowing problems. The main goal of this study was to evaluate an automatic tool based on voice analysis to support medical detection of dysphagia.

Methods: An automatic voice analysis system has been developed. Prior to (basal) and immediately following (viscosity) swallowing liquids of varying viscosity and volume, individuals with Parkinson Disease were required to produce the same test word. The acoustic features (linear and non-linear) of this word were then analyzed with regard to specific situations by standard and Machine Learning methods.

Results: The results indicated a high degree of accuracy in detecting voice associated with basal and viscosity states.

Conclusion: Thus, while the gold standard of dysphagia diagnosis continues to involve video-fluoroscopy analysis, the consideration of voice analysis may prove to be a far simpler and less invasive approach to diagnosis by advanced voice features.",
  address="Taylor and Francis Ltd.",
  chapter="140606",
  doi="10.1080/2050571X.2017.1369017",
  howpublished="print",
  institution="Taylor and Francis Ltd.",
  number="2",
  volume="21",
  year="2018",
  month="april",
  pages="86--89",
  publisher="Taylor and Francis Ltd.",
  type="journal article"
}