Inter-rater and Intra-rater Reliability of Glottal Image Capture: A Mobile Application to Quantify Vocal Fold Bowing.

Abstract

Age-related vocal atrophy (ARVA) causes vocal fold bowing, impacting communication and quality of life. The assessment of vocal fold bowing is largely subjective. Glottal Image Capture (GlottIC) is a new mobile application that helps quantify vocal fold bowing. We aim through this study to assess its reliability, compare it to manual calculation method, and compare differences between visual-perceptual bowing severity ratings. Ten raters independently quantified Bowing Index (BI) using GlottIC from 10 videostroboscopic images among individuals with ARVA. There was 100% duplication of images to facilitate intra-rater reliability analyses using Pearson's correlation. Inter-rater reliability was quantified using Intraclass Correlation Coefficient (ICC) for experienced and novice raters. The correlation between manual calculations using ImageJ and GlottIC was analyzed. The intra-rater reliability for total BI was strong (r = 0.822, p < 0.001). The inter-rater reliability for BI, calculated using ICC, was (ICC = 0.720; 95% CI: 0.579-0.852), indicating good consistency among the raters. Experts had higher ICC (ICC = 0.808; 95% CI: 0.678-0.906) compared to novice raters (ICC = 0.651; 95% CI: 0.468-0.816). There was a positive correlation between GlottIC and manual BI (r = 0.811, p < 0.001). As the BI increased, the bowing severity, based on visual-perceptual ratings, also increased (p < 0.001). GlottIC is a reliable mobile application that can quantify vocal fold bowing in patients with ARVA with high intra- and inter-rater reliability. GlottIC BI measurments are highly correlated with manual BI and visual-perceptual ratings of bowing severity. Further improvements in reliability may be achieved with more robust rater training and automated technologies. Level 3 Laryngoscope, 2024.