Adductory Vocal Fold Kinematic Trajectories During Conventional Versus High-Speed Videoendoscopy

Abstract

Objective Prephonatory vocal fold angle trajectories may supply useful information about the laryngeal system but were examined in previous studies using sigmoidal curves fit to data collected at 30 frames per second (fps). Here, high-speed videoendoscopy (HSV) was used to investigate the impacts of video frame rate and sigmoidal fitting strategy on vocal fold adductory patterns for voicing onsets. Method Twenty-five participants with healthy voices performed /ifi/ sequences under flexible nasendoscopy at 1,000 fps. Glottic angles were extracted during adduction for voicing onset; resulting vocal fold trajectories (i.e., changes in glottic angle over time) were down-sampled to simulate different frame rate conditions (30–1,000 fps). Vocal fold adduction data were fit with asymmetric sigmoids using 5 fitting strategies with varying parameter restrictions. Adduction trajectories and maximum adduction velocities were compared between the fits and the actual HSV data. Adduction trajectory errors between HSV data and fits were evaluated using root-mean-square error and maximum angular velocity error. Results Simulated data were generally well fit by sigmoid models; however, when compared to the actual 1,000-fps data, sigmoid fits were found to overestimate maximum angle velocities. Errors decreased as frame rate increased, reaching a plateau by 120 fps. Conclusion In healthy adults, vocal fold kinematic behavior during adduction is generally sigmoidal, although such fits can produce substantial errors when data are acquired at frame rates lower than 120 fps.

An Open‐Source Computer Vision Tool for Automated Vocal Fold Tracking From Videoendoscopy

“…The authors reported an intraclass correlation coefficient of 0.85 between expert markings and their algorithm estimations and noted that manual marking of the AGA was often necessary due to suboptimal algorithm performance. 16 Computer vision is a branch of artificial intelligence (AI) that enables automated localization of anatomic landmarks from digital media. Machine learning is a subset of AI that employs task-specific predictive algorithms programmed to self-adjust in response to training data.…”

“…These reports, however, did not report comparison of algorithm estimations against expert markings. Diaz‐Cadiz et al 16 characterized the impact of varying camera frame rates on the estimation of vocal fold dynamics using traditional image‐processing techniques. The authors reported an intraclass correlation coefficient of 0.85 between expert markings and their algorithm estimations and noted that manual marking of the AGA was often necessary due to suboptimal algorithm performance 16 …”

Acoustic Identification of the Voicing Boundary during Intervocalic Offsets and Onsets Based on Vocal Fold Vibratory Measures

“…In brief, technicians were first trained to measure glottic angles (extending from the anterior commissure along the medial vocal fold edge to the vocal process) from images obtained during a flexible nasendoscopic procedure using a halogen light source and acquired at a conventional framerate of 30 frames per second. Technicians were required to meet two-way mixed-effects intraclass correlation coefficients (ICC) for consistency of agreement ≥0.80 when compared to glottic angle markings made previously by a gold-standard technician [38]. The average reliability for the nine technicians was ICC(3,1) = 0.89 (SD = 0.01, range = 0.88-0.91).…”

“…The nine technicians then completed training to use a semi-automated glottic angle tracking algorithm, as described in detail in Diaz-Cadiz et al [38]. Using this algorithm, the technicians were trained to use time-aligned microphone signal and video frames captured during an /ifi/ utterance to semi-automatically estimate the glottic angle over time.…”

A Deep Learning Enhanced Novel Software Tool for Laryngeal Dynamics Analysis

“…The evaluation in videostroboscopy is typically performed subjectively and hence naturally relies on the expertise of the clinician rather than measurable, quantitative parameters (Olthoff et al, 2007). Although stroboscopy principally allows for glottis segmentation and hence potential computation of quantitative vibratory parameters (Woo, 2020), HSV seems to be better suited for objective assessment and analysis of vocal fold vibrations as many studies have shown (Bohr et al, 2014;Diaz-Cadiz et al, 2019;Yamauchi et al, 2017). A common way to quantify the vocal fold oscillation in HSV is analyzing the changing glottal area over time (Deliyski et al, 2008).…”