Estimating Vocal Fold Contact Pressure from Raw Laryngeal High-Speed Videoendoscopy Using a Hertz Contact Model

Abstract

The development of trauma-induced lesions of the vocal folds (VFs) has been linked to a high collision pressure on the VF surface. However, there are no direct methods for the clinical assessment of VF collision, thus limiting the objective assessment of these disorders. In this study, we develop a video processing technique to directly quantify the mechanical impact of the VFs using solely laryngeal kinematic data. The technique is based on an edge tracking framework that estimates the kinematic sequence of each VF edge with a Kalman filter approach and a Hertzian impact model to predict the contact force during the collision. The proposed formulation overcomes several limitations of prior efforts since it uses a more relevant VF contact geometry, it does not require calibrated physical dimensions, it is normalized by the tissue properties, and it applies a correction factor for using a superior view only. The proposed approach is validated against numerical models, silicone vocal fold models, and prior studies. A case study with high-speed videoendoscopy recordings provides initial insights between the sound pressure level and contact pressure. Thus, the proposed method has a high potential in clinical practice and could also be adapted to operate with laryngeal stroboscopic systems.

Keywords: Hertz impact; biomechanical modeling; contact pressure; endoscopy; high-speed video; laryngeal high-speed videoendoscopy; tissue modeling; vocal folds.

Figure 1.

A schematic of a vocal fold (VF) motion sequence in a typical videoendoscopy recording. The colored lines represent the edge estimation of each fold (left in red and right in blue). The fictitious edge overlap during collision shows the estimated penetration depth δ_c and contact length L_c.

Figure 2.

Edge detection example using a HSV recording. Representative steps of the detection process and a mid portion kymography of the glottis are shown. Gradient information is used (first row) to find left and right VF borders. VF edge points are missing during collision (second row). Edge-based kymogram (third row) and its temporally averaged version (fourth row) are also shown. Note that temporal averaging reduces the errors incurred by the edge detection, but it does not complete the trajectory in closing phases.

Figure 3.

Coefficient tracking using the Kalman Filter(KF) during closure. Quadratic curves are employed (p = 2). The estimates of Θ_k are converted to standard quadratic coefficients as ${(a,b,c)}_k={\left(1,-\left(y_a+yb\right),x_a+y_a{y}_b\right)}_k{\theta }_{0,k}$ and are plotted across time. The polynomial coefficients of each fold are shown before and after the KF edge tracking module, with the ratio λ_k and uncertainty factor ρ_k (only the uncertainty of the right side is shown for simplicity). The VF description fails when collision occurs, making registered coefficients not valid at certain times. The KF completes the temporal sequence of the fitted model by making predictions of its location during the collision phases (high ρ_k values).

Figure 4.

KF Edge tracking result. Three representative instants of the closure and a midline kymogram are shown. An overlap between VF is now visible during contact phases despite the lost of detected points in the detection stage.

Figure 5.

Geometry and dimensions of the synthetic vocal fold model. All reported dimensions have units of cm.

Figure 6.

A schematic of the experimental flow facility. All dimensions shown are in mm.

Figure 7.

Right side kymograms with upper and lower mass displacements from Modified Body Cover Model (MBCM) and CPA trajectory predictions. Two particular simulations from (a) TEST1 and (b) TEST2 at Ps = 1500 Pa are shown. CPA-detected and -tracked trajectories were obtained only by processing the videos. CPA-detected points were used to reproduce the lower mass penetration in both cases, considering their kinematics.

Figure 8.

Numerical validations for both complete (top row; TEST1) and incomplete (bottom row; TEST2) glottal closure scenarios. Apparent penetration (left column) and normalized pressure (right column) are presented for the reference model MBCM, the idealized HERTZ contact accounting for the superior and inferior tissue, and the proposed CPA scheme using the superior view only. Note that subglottal pressures below 1000 Pa were not sufficient to cause VF contact under incomplete glottal closure scenarios.

Figure 9.

Hertzian pressure estimations for in vivo HSV recordings. Glottal area, penetration, and contact area estimates for soft (solid line), normal (dash line), and loud (dash-dot line) gestures are shown. Increasing impact slopes of glottal area are observed with increasing loudness. Increasing penetration depth and estimated normalized contact pressure are observed with louder gestures.

Figure 10.

Validation of the proposed Hertz contact model (red lines) against published data (blue lines). For ease of comparison, the time axis was normalized to maintain a contact duration of 1 ms.