Quantitative Endoscopy for Airway Evaluation: A potential Clinical Paradigm-Shifting Technology

Project Summary:

Endoscopy is used in the daily practice of clinicians for a multitude of indications and has become an increasingly employed tool in the evaluation of upper airway (UA) disorders. Endoscopic technology provides visualization to assist in the diagnosis and treatment response assessment, but at this time it does not allow for rapid quantification of UA diameter or cross‐sectional area (CSA). This transformative project will develop quantitative endoscopy methods based on an entirely new paradigm, namely the use of artificial intelligence (AI) to estimate depth maps.

In a pilot study, this project team demonstrated that the airspace CSAs of 3-dimensional (3D) models built from computed tomography (CT) can be accurately estimated from depth maps. In the pilot study, the depth maps could be generated because the 3D geometry was known from the CT scans. In a follow-up unpublished study, the team demonstrated the feasibility of training an AI algorithm to predict depth maps of 3D models.

This project aims to develop and validate AI methods to compute depth maps from clinical endoscopy. The airspace CSAs estimated with the novel quantitative endoscopy technique will be validated with in vivo measurements of the nasal cavity minimal CSA using acoustic rhinometry. Quantitative endoscopy will revolutionize surgical planning for obstructive UA disorders by providing a tool to rapidly quantify airway size at the point of care. In particular, obstructive sleep apnea (OSA) is a common disease in Wisconsin. Today, surgical planning for OSA is based on subjective scoring of endoscopic observations without quantifying UA collapsibility, which leads to poor surgical outcomes in greater than 50% of patients. Quantitative endoscopy is a paradigm-shifting technology that will allow quantification of the collapsibility of each pharyngeal structure (tongue, soft palate, lateral pharyngeal walls, and epiglottis), thus providing objective criteria to identify the optimal intervention for each patient.