AI Detects Palatoglossal Airspace on Panoramic Radiographs to Assess Image Quality

An AI model (YOLOv11n) detected palatoglossal airspace on panoramic radiographs with 89.3% accuracy, balanced sensitivity and specificity (89.47%), and nearly perfect agreement (Kappa=1.0) in 456 radiographs, offering a tool to flag compromised image quality before diagnostic in…

• In 456 panoramic radiographs, the AI model achieved 89.3% accuracy, 89.47% sensitivity, and 89.47% specificity for detecting palatoglossal airspace, indicating tongue malposition.
• The model showed perfect agreement with the reference standard (Kappa = 1.0), suggesting robust classification.
• The algorithm is publicly accessible via a website for instant technical analysis of panoramic images.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.