Automatic Detection and Segmentation of Tooth Cracks Based on Improved Mask R-CNN

Area of Science:

• Artificial Intelligence in Dentistry
• Medical Image Analysis
• Dental Diagnostics

Background:

• Cracked teeth pose a diagnostic challenge due to subtle and irregular features.
• Early detection is vital to prevent further dental damage and complications.

Purpose of the Study:

• To develop an automated system for accurate cracked tooth detection and segmentation.
• To improve upon existing methods for identifying subtle dental fractures.

Main Methods:

• An improved Mask R-CNN instance segmentation network was utilized.
• ResNeXt-50(32×4d) backbone and a Crack Feature Enhancement Module (CFEM) were incorporated.
• A redesigned mask head with encoder-decoder structure and dynamic snake convolutions was implemented.

Main Results:

• The proposed method demonstrated superior performance compared to state-of-the-art techniques.
• Experiments on real intraoral images confirmed the model's effectiveness in detecting tooth cracks.
• The system achieved more accurate and earlier detection of cracked teeth.

Conclusions:

• The developed AI model offers a significant advancement in diagnosing cracked teeth.
• This technology facilitates timely interventions, reduces invasive treatments, and preserves tooth structure.
• The availability of code and datasets promotes further research and application in dental diagnostics.