Semi-automatic tracking of hyolaryngeal coordinates in videofluoroscopic swallowing studies

Abstract

The gold standard for assessing dysphagia in the clinical setting is the videofluoroscopic swallowing study. Detailed, quantitative analysis of swallowing videos has significant potential to improve our understanding of healthy and pathological swallows. A key limitation of such analyses is that manual annotation of anatomical landmarks on video frames is tedious, time-consuming and potentially unreliable. In this study, we demonstrate a computer-assisted workflow that uses the Kanade–Lucas–Tomasi tracker to semi-automate the process of collecting coordinate data throughout a swallow cycle. We tested the workflow against eight videofluorographs of normal swallow subjects obtained from a variety of sources. Analysis of results showed that this workflow can facilitate the process of accurately determining a variety of hyolaryngeal kinematic variables and generate coordinates useful for biomechanical simulation and morphometric analysis of swallowing.

Keywords::

• applications of imaging and visualization
• image processing and analysis
• imaging and visualization in biomechanics
• data processing and analysis