Estimation of Nasal Airway Cross-sectional Area From Endoscopy Using Depth Maps: A Proof-of-Concept Study

Area of Science:

• Medical imaging
• Otorhinolaryngology
• Pulmonary medicine

Background:

• Endoscopy is vital for diagnosing obstructive airway diseases.
• Current endoscopy lacks airspace cross-sectional area (CSA) quantification.
• Virtual endoscopy offers potential for quantitative analysis.

Purpose of the Study:

• To test if CSAs can be accurately estimated from virtual endoscopy depth maps.
• To validate a software tool for airway CSA measurement.
• To assess the potential for AI in clinical endoscopy quantification.

Main Methods:

• Created virtual endoscopy and depth map videos from 3D CT models of 30 subjects.
• Developed software to measure airway perimeter and estimate CSA from depth maps.
• Two otolaryngologists measured nasopharynx and nasal valve CSAs, comparing to true 3D model values.

Main Results:

• Nasopharynx CSA estimation showed low median percent error (3.7%-4.6%).
• Nasal valve minimal CSA (mCSA) estimation had higher median percent error (22.7%-33.6%).
• The tool accurately identified nasal valve stenosis with high sensitivity, specificity, and accuracy.

Conclusions:

• Airway CSAs can be accurately estimated from depth maps derived from 3D models.
• This technique shows promise for quantitative analysis in virtual endoscopy.
• Future AI development could enable CSA quantification from clinical endoscopies.