| dc.description.abstract | Unilateral vocal fold paralysis (UVFP) is a condition in which one side of the vocal fold loses its adduction function and thus the vocal fold cannot vibrate normally, leading to difficulties in voice production. In type-1 thyroplasty, an implant is introduced through the thyroid cartilage to medialize the paralyzed vocal fold and restore voice. To improve implant design, our primary focus is to develop a high-fidelity fluid-structure interaction (FSI) model of the vocal fold undergoing the surgery, and to validate the model using integrated experimental studies of rabbit larynges.
First, we performed 3D FSI simulations of healthy phonation, where the two sides of the vocal fold were medialized symmetrically. By incorporating the subject-specific anatomical features reconstructed from magnetic resonance scans, as well as individualized tissue properties estimated using a simple flow model, the high-fidelity model was able to capture the subject-specific vibratory characteristics that matched the in vivo phonation test.
Next, we conducted an integrated experimental and computational study of an ex vivo rabbit larynx at both the UVFP and type-1 thyroplasty conditions. The model was built upon the pre-operative scan and the simulation consisted of two steps: 1) a finite-element method simulation of the vocal fold adduction to model the medialization of the implant; 2) an FSI simulation with the implant incorporated to model the vibration of the vocal fold. The results from both steps agree with experiment data, i.e., post-operative scan and high-speed imaging, thus showing successful model validation.
Finally, we employed the computational model to optimize implant position and depth. The degree of static displacement due to implant medialization and the vibration amplitude were used for implant assessment. The results show that the optimal implant position for vibration differed from that for maximal displacement, which highlights the need for FSI modeling to predict the implant's comprehensive effects. The model prediction is generally aligned with previous experimental studies but also shows that optimal implant depends on subject-specific features. Thus, a computational modeling-based tool would be useful for pre-surgical planning. | |
