Comparison of AI-Powered Tools for CBCT-Based Mandibular Incisive Canal Segmentation: A Validation Study

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Accurate identification of the mandibular incisive canal (MIC) is crucial for safe anterior dental implant placement.
• Traditional methods for MIC identification on cone beam computed tomography (CBCT) scans can be challenging and time-consuming.

Purpose of the Study:

• To validate an enhanced AI-driven model for automated MIC segmentation on CBCT scans.
• To compare the accuracy and time efficiency of the enhanced AI model against human experts and a prior AI model.

Main Methods:

• Developed an enhanced AI model using expert-optimized MIC segmentation on 100 CBCT scans.
• Tested the model's performance against human experts and a previous AI model on 40 CBCT scans.
• Evaluated metrics including IoU, DSC, recall, precision, accuracy, and RMSE, alongside time efficiency.

Main Results:

• The enhanced AI model achieved high performance metrics: 93% IoU, 93% DSC, 94% recall, 93% precision, 99% accuracy, and 0.23 mm RMSE.
• Demonstrated significantly superior performance compared to the previous AI model and manual segmentation for key metrics (p < 0.0001).
• Completed segmentation in 17.6 seconds, proving 125 times faster than manual segmentation (p < 0.0001).

Conclusions:

• The enhanced AI model offers accurate and efficient automated MIC segmentation.
• Its performance surpasses both human expert and previous AI model segmentation.
• This AI tool holds significant potential for improving pre-surgical planning in anterior implantology.