Physical parameter estimation from porcine ex vivo vocal fold dynamics in an inverse problem framework

Gómez, Pablo; Schützenberger, Anne; Kniesburges, Stefan; Bohr, Christopher; Döllinger, Michael

doi:10.1007/s10237-017-0992-5

Cited by 15 publications

(41 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to their inherent simplicity and the versatility of the models, they should have a potential role in optimization or deep learning algorithms. On the basis of endoscopic high-speed video footage from patients, the model can be applied to estimate the subglottal oscillation threshold pressure similar to the methodology reported by Gomez et al [47,48]. Assuming a high accuracy of the estimation, the methods make an important parameter available for evaluating the effort that a patient has to make to phonate.…”

Section: Discussionmentioning

confidence: 99%

Geometry of the Vocal Tract and Properties of Phonation near Threshold: Calculations and Measurements

et al. 2019

Self Cite

View full text Add to dashboard Cite

In voice research, analytically-based models are efficient tools to investigate the basic physical mechanisms of phonation. Calculations based on lumped element models describe the effects of the air in the vocal tract upon threshold pressure (Pth) by its inertance. The latter depends on the geometrical boundary conditions prescribed by the vocal tract length (directly) and its cross-sectional area (inversely). Using Titze’s surface wave model (SWM) to account for the properties of the vocal folds, the influence of the vocal tract inertia is examined by two sets of calculations in combination with experiments that apply silicone-based vocal folds. In the first set, a vocal tract is constructed whose cross-sectional area is adjustable from 2.7 cm2 to 11.7 cm2. In the second set, the length of the vocal tract is varied from 4.0 cm to 59.0 cm. For both sets, the pressure and frequency data are collected and compared with calculations based on the SWM. In most cases, the measurements support the calculations; hence, the model is suited to describe and predict basic mechanisms of phonation and the inertial effects caused by a vocal tract.

show abstract

Section: Discussionmentioning

confidence: 99%

Geometry of the Vocal Tract and Properties of Phonation near Threshold: Calculations and Measurements

et al. 2019

Self Cite

View full text Add to dashboard Cite

show abstract

“…While the precise procedure used in the inverse problem formulation of previous studies varies [10], [12], [15], [43], the general idea remains similar. Given some noise-affected data of real vocal fold oscillations \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$y$ \end{document}, we want to find a configuration \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$q$ \end{document} for a numerical model \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$G$ \end{document} such that \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}\begin{equation*} y \approx G(q).\tag{1}\end{equation*} \end{document} If the problem is solved successfully, i.e.…”

Section: Methods and Proceduresmentioning

confidence: 99%

“…This approach was employed in a previous study [10] to estimate the subglottal pressure present in ex vivo experiments using porcine larynges. In this study, we are relying on the same numerical model used to generate labeled training data and instead train a neural network to estimate the subglottal pressure.…”

Section: Methods and Proceduresmentioning

confidence: 99%

“…If the numerical model is physically sound and the optimization successful, it is possible to infer the parameters of the real vocal folds. First suggested by Döllinger et al [12] this approach has been successfully used to, e.g., estimate the subglottal pressure [10] or study disorders such as unilateral vocal fold paralysis [14].…”

Section: Related Workmentioning

confidence: 99%

“…We demonstrated in a previous study [10] that it is possible to estimate the subglottal pressure in a numerical inverse problem. However, the previous results indicate that the optimization would benefit from a more sophisticated model, which is not computationally feasible.…”

Section: Introductionmentioning

confidence: 98%

See 2 more Smart Citations

Laryngeal Pressure Estimation With a Recurrent Neural Network

Gómez

Schützenberger

Semmler

et al. 2019

IEEE J. Transl. Eng. Health Med.

Self Cite

View full text Add to dashboard Cite

Quantifying the physical parameters of voice production is essential for understanding the process of phonation and can aid in voice research and diagnosis. As an alternative to invasive measurements, they can be estimated by formulating an inverse problem using a numerical forward model. However, high-fidelity numerical models are often computationally too expensive for this. This paper presents a novel approach to train a long short-term memory network to estimate the subglottal pressure in the larynx at massively reduced computational cost using solely synthetic training data. We train the network on synthetic data from a numerical two-mass model and validate it on experimental data from 288 high-speed ex vivo video recordings of porcine vocal folds from a previous study. The training requires significantly fewer model evaluations compared with the previous optimization approach. On the test set, we maintain a comparable performance of 21.2% versus previous 17.7% mean absolute percentage error in estimating the subglottal pressure. The evaluation of one sample requires a vanishingly small amount of computation time. The presented approach is able to maintain estimation accuracy of the subglottal pressure at significantly reduced computational cost. The methodology is likely transferable to estimate other parameters and training with other numerical models. This improvement should allow the adoption of more sophisticated, high-fidelity numerical models of the larynx. The vast speedup is a critical step to enable a future clinical application and knowledge of parameters such as the subglottal pressure will aid in diagnosis and treatment selection.

show abstract