Micro-Computed Tomography-Guided Artificial Intelligence for Pulp Cavity and Tooth Segmentation on Cone-beam Computed...

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Cone-beam computed tomographic (CBCT) imaging is crucial for dental diagnostics.
• Accurate segmentation of tooth and pulp cavity in CBCT images is challenging.
• Existing segmentation methods may lack precision and automation.

Purpose of the Study:

• To develop and evaluate a novel data pipeline for automated tooth and pulp cavity segmentation using micro-CT data.
• To train a U-Net network for enhanced segmentation accuracy on CBCT images.
• To compare the performance of the proposed micro-CT data pipeline against manual segmentation.

Main Methods:

• Collected CBCT and micro-CT data from 30 teeth.
• Processed CBCT data into high-resolution images.
• Utilized a micro-CT data pipeline for U-Net training (experimental group) versus manual segmentation (control group).
• Evaluated segmentation using Dice similarity coefficient, precision, recall, and surface distance metrics.

Main Results:

• The micro-CT data pipeline achieved superior segmentation accuracy for both tooth and pulp cavity compared to manual segmentation.
• Quantitative metrics (Dice, precision, recall, distances) demonstrated significant improvements with the experimental group.
• Morphologic analysis confirmed better segmentation quality with the proposed method.

Conclusions:

• The proposed micro-CT data pipeline offers an automatic and accurate approach for tooth and pulp cavity segmentation on CBCT images.
• This method holds potential for improving research and clinical applications in dentistry.
• The U-Net network, trained with the novel pipeline, shows promise for precise dental image analysis.