A Deep Learning Model for Idiopathic Osteosclerosis Detection on Panoramic Radiographs

Area of Science:

• Dentistry
• Radiology
• Artificial Intelligence

Background:

• Idiopathic osteosclerosis (IO) is a condition often identified on panoramic radiographs.
• Accurate and efficient detection of IO is crucial for dental diagnostics.

Purpose of the Study:

• To develop an artificial intelligence (AI) system for automated detection of idiopathic osteosclerosis (IO) on panoramic radiographs.
• To create a simple and routine evaluation tool for IO detection.

Main Methods:

• A deep learning model (GoogLeNet Inception v2) was trained using 493 anonymized panoramic radiographs.
• The AI system, CranioCatch, was developed using TensorFlow for IO detection.
• Model performance was evaluated using a confusion matrix.

Main Results:

• The AI model achieved high accuracy in detecting IOs on test images.
• Sensitivity, precision, and F-measure values were 0.88, 0.83, and 0.86, respectively.
• The system accurately identified 50 out of 57 IOs in 52 test images.

Conclusions:

• Deep learning-based AI algorithms can accurately detect IOs on panoramic radiographs.
• AI systems have the potential to significantly reduce dentists' diagnostic workload.
• The developed AI system offers a promising tool for routine dental evaluations.