Deep learning-based tooth segmentation methods in medical imaging: A review.

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Dentistry

Background:

• Deep learning, particularly convolutional neural networks (CNNs) and Transformers, has significantly advanced medical image analysis.
• Tooth segmentation is crucial for clinical dental assessments, pathology diagnosis, and surgical planning.
• Existing deep learning models like U-Net, Mask R-CNN, and SETR provide foundational frameworks for tooth segmentation.

Purpose of the Study:

• To review deep learning methodologies for tooth segmentation across various dental imaging modalities.
• To discuss performance-enhancing techniques and identify limitations in current tooth segmentation research.
• To provide insights for future research and facilitate broader clinical adoption of automated tooth segmentation.

Main Methods:

• Review of deep learning techniques including CNNs and Transformers for tooth segmentation.
• Analysis of models applied to dental panoramic radiographs (DPRs), cone-beam computed tomography (CBCT) images, and intraoral scan (IOS) models.
• Discussion of enhancement and optimization modules for improved segmentation performance.

Main Results:

• Deep learning models have shown significant progress in deriving tooth feature maps.
• Various architectures and enhancement modules have been proposed to improve segmentation accuracy.
• Challenges such as data annotation and model generalization persist, impacting widespread clinical use.

Conclusions:

• Deep learning offers powerful tools for automated tooth segmentation in dentistry.
• Further research is needed to address current limitations and enhance model robustness.
• Improved tooth segmentation can significantly aid clinical decision-making and patient care.