A reliable deep-learning-based method for alveolar bone quantification using a murine model of periodontitis and...

Area of Science:

• Biomedical Engineering
• Dental Research
• Artificial Intelligence in Medicine

Background:

• Periodontitis leads to alveolar bone loss, a critical factor in tooth support.
• Micro-computed tomography (µCT) is essential for visualizing and quantifying bone structure.
• Manual analysis of µCT data for alveolar bone is time-consuming and requires specialized expertise.

Purpose of the Study:

• To develop an automated deep-learning segmentation model for analyzing alveolar bone in a mouse model of periodontitis.
• To enable researchers to easily examine alveolar bone from µCT data without prior machine learning knowledge.
• To accurately measure alveolar bone volume (BV) and bone mineral density (BMD) while excluding teeth.

Main Methods:

• Ligature-induced experimental periodontitis was established in mice.
• Maxillary bone samples were harvested at various time points (4, 7, 9, 14 days) and scanned using µCT.
• A 3D deep learning model based on the U-Net architecture was developed using Dragonfly software for image segmentation.

Main Results:

• The AI model achieved over 98% accuracy in segmenting alveolar bone from µCT data.
• Alveolar bone volume (BV) on the ligature side showed a progressive decrease, reaching a minimum on day 9.
• Bone mineral density (BMD) on the ligature side decreased significantly over time, with a 21.1% loss by day 14.

Conclusions:

• An accessible, downloadable AI model was developed for the automatic segmentation and analysis of mouse alveolar bone.
• The model accurately quantifies key metrics like BV, BMD, and trabecular bone thickness, excluding teeth.
• This user-friendly tool enhances the potential of AI in diagnosing and treating oral diseases.