AI-Base Dysphagia Prediction using Panoramic

A Basic Study for Predicting Dysphagia in Panoramic X-ray Images Using Artificial Intelligence (AI)—Part 1: Determining Evaluation Factors and Cutoff Levels

Authors: Yukiko Matsuda, Emi Ito, Migiwa Kuroda and Kazuyuki Araki

---

Summary
This study investigated whether panoramic radiographs can be used to detect indicators of dysphagia using artificial intelligence. Seventy-seven patients who underwent both panoramic radiography and videofluorographic (VF) swallowing studies were analyzed. Multiple anatomical and positional features were evaluated, including the vertical and horizontal positions of the hyoid bone, the distance between the tongue and palate, and the width of the tongue. Logistic regression analysis revealed that only the v... The study established a cutoff level based on ROC analysis (AUC = 0.72), determining that when the hyoid bone is positioned below the mandibular border line, the risk of dysphagia increases significantly. The authors propose developing an AI model for automated screening of dysphagia risk using panoramic radiographs.

DOI: https://doi.org/10.3390/ijerph19084529

---

Key Words

AI in Dentistry, Panoramic Radiography, Dysphagia, Hyoid Bone, Deep Learning, Radiographic Analysis, Cutoff Level, ROC Curve

---

Extracted Data

• Year: 2022
• Modality: Panoramic
• Dataset: 77 patients
• Dataset Split: Dysphagia (+) = 47 / Dysphagia (−) = 30
• Network Architecture: Logistic Regression
• Metrics: ROC AUC = 0.72; Cutoff level = Grade 0 (hyoid below mandibular border line)
• AP – Strategy: No information
• AP – Professional Qty: 2
• AP – Supervisor Presence: No information
• AP – Experience Level: No information
• AP – Expertise Area: Oral and Maxillofacial Radiology
• AP – Tool or System: JMP Pro v16; SPSS v27 (Statistical analysis)
• Task: Classification
• Project Objective: Establish objective parameters to support the future development of AI for dysphagia screening

---

Clinical Relevance

• Clinical significance: This study demonstrates how conventional dental panoramic radiographs can contribute to systemic health evaluation beyond oral conditions.
• Scientific contribution: By identifying the vertical hyoid bone position as a radiographic marker for dysphagia, the authors provide a foundation for AI-assisted early detection of swallowing disorders.
• Practical impact: The integration of panoramic imaging and AI has potential to enhance patient safety in dental procedures, assist in geriatric care, and enable preventive screening for oral frailty and aspiration risk.

This work represents one of the first structured attempts to quantify dysphagia risk using dental imaging — bridging radiology, geriatrics, and artificial intelligence.

---

Related Video

▶️ Watch on YouTube