Automatic tooth periodontal ligament segmentation of cone beam computed tomography based on instance segmentation...

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Periodontal ligament (PDL) morphology is crucial for various dental interventions.
• Accurate PDL segmentation is essential for effective diagnosis and treatment planning.

Purpose of the Study:

• To develop and evaluate a deep learning (DL) algorithm for automatic PDL segmentation in cone-beam computed tomography (CBCT) images.
• To assess the accuracy and efficiency of AI-driven PDL segmentation for clinical applications.

Main Methods:

• A retrospective study utilizing 1734 axial CBCT images from 389 patients.
• Development of a Mask R-CNN based instance segmentation model for automatic PDL segmentation.
• Model training used 'teeth' and 'alveolar bone' labels, defining PDL as their overlapping region.

Main Results:

• High qualitative accuracy achieved for incisors, canines, premolars, wisdom teeth, and implants (100%).
• Molars showed 96.4% segmentation accuracy.
• Quantitative metrics demonstrated mean Intersection over Union (mIoU) of 0.667 ± 0.015 and mean Dice Similarity Coefficient (mDSC) of 0.799 ± 0.015.

Conclusions:

• The AI-driven approach offers a novel method for automatic PDL segmentation on CBCT.
• This technology has the potential to enable chair-side measurements, improving efficiency and accuracy for periodontists, orthodontists, prosthodontists, and implantologists.