Image Resolution's Impact on Artificial Intelligence & Human Accuracy in Full-Mouth Radiographic Analysis.

Area of Science:

• Dentistry
• Radiology
• Artificial Intelligence

Background:

• Artificial intelligence (AI) is increasingly utilized for dental radiographic interpretation.
• The impact of image resolution on AI diagnostic accuracy compared to human experts is not well-established.

Purpose of the Study:

• To evaluate how medium- versus high-resolution full-mouth radiographs affect the diagnostic performance of an AI system and experienced clinicians.
• To compare diagnostic accuracy metrics between different image resolutions for AI and human interpretation.

Main Methods:

• A retrospective study analyzed 200 full-mouth series radiographs, divided into medium-resolution (96-300 dpi) and high-resolution (≥720 dpi) groups.
• Three human examiners and an AI system (Diagnocat) assessed six pathological conditions, with reference standard based on inter-examiner agreement.
• Diagnostic metrics (sensitivity, specificity, accuracy, F1-score) and reliability (Cohen's kappa) were calculated; differences between resolutions were statistically tested.

Main Results:

• High-resolution imaging significantly improved AI diagnostic accuracy for caries, furcation involvement, dental calculus, and missing teeth (P < .001).
• AI accuracy ranged from 75.3% to 99.1%, with high specificity (73.5%-99.8%) but variable sensitivity (0.0%-93.6%).
• Human inter-examiner agreement varied from moderate to substantial (κ = 0.27-0.87); AI-gold standard agreement improved with higher resolution.

Conclusions:

• High-resolution radiographs enhance diagnostic accuracy for both AI and human dental interpretations.
• AI demonstrated clinically acceptable performance, though sensitivity varied across different pathologies.
• Optimal image quality is essential for improving diagnostic confidence and supporting AI-assisted decision-making in dentistry.