Application of an AI-Based Model for Non-Invasive Sonographic Assessment for Injection Laryngoplasty.

Abstract

Objective: Hyaluronic acid (HA) can be degraded over time. However, the persistence of the effects after injection laryngoplasty (IL) for unilateral vocal fold paralysis (UVFP) has been observed. The relation between HA residue and clinical voice outcomes is undetermined. We aimed to develop a non-invasive method with clinical applicability for objective evaluation of the temporal change in HA volume using artificial intelligence-based ultrasonographic assessment. Furthermore, the volume outcome and its relation to voice outcomes were investigated.

Study design: A prospective cohort study.

Setting: A voice clinic at a tertiary medical center.

Methods: Adult patients who underwent HA IL for UVFP were recruited. The Voice Handicap Index-10 questionnaire, the grade, roughness, breathiness, asthenia, strain scale, and the normalized glottal gap were evaluated before IL. Patients received ultrasonography for the identification of HA and estimation of volume based on Temporal Convolutional Segmentation Network, and also voice outcome assessments at 2 weeks, and 2 and 6 months after IL.

Results: Contours of HA were well-described, and volumes were estimated accordingly. Clinical voice outcomes were significantly improved postoperatively and remained stationary. Contrarily, residual HA volume decreased significantly across time points (1.21 ± 0.53 at 2 weeks, 0.63 ± 0.35 at 2 months, P < .001; 0.32 ± 0.25 at 6 months, P < .001). Simple linear regression showed no significant relationship between the remaining volume at 2 months and patient age or duration of symptoms (P = .67).

Conclusion: Volume change of injected HA over time was estimated non-invasively. The prolonged voice improvement, longer than HA longevity, was demonstrated objectively. The knowledge on HA longevity and duration of benefit provides perspectives beneficial to counseling patients and future treatment planning.