Segmentation of periapical lesions with automatic deep learning on panoramic radiographs: an artificial intelligence...

Area of Science:

• Dentistry
• Radiology
• Artificial Intelligence

Background:

• Periapical periodontitis commonly presents as radiolucent lesions on panoramic radiographs.
• Accurate detection of these lesions is crucial for timely diagnosis and treatment.

Purpose of the Study:

• To evaluate the diagnostic accuracy of a U²-Net based artificial intelligence (AI) model for detecting periapical lesions on dental panoramic radiographs.
• To assess the AI model's potential to assist clinicians in diagnosis and enhance workflow efficiency.

Main Methods:

• A retrospective analysis of 400 panoramic radiographs containing periapical radiolucencies.
• Manual labeling of 780 periapical radiolucencies by two independent examiners.
• Training a U²-Net architecture AI model using a deep supervision algorithm.

Main Results:

• The AI model achieved a Dice score of 0.8 on the validation set.
• The model demonstrated a precision of 0.82, recall of 0.77, and F1-score of 0.8 on the test set.
• The U²-Net based AI model accurately diagnosed periapical lesions.

Conclusions:

• AI models based on U²-Net architecture show high accuracy in diagnosing periapical lesions from panoramic radiographs.
• AI tools offer promising applications to support dentists in diagnosing periapical radiolucencies and planning procedures.
• Larger datasets are recommended for future studies to further enhance AI diagnostic accuracy.