Intelligent Virtual Dental Implant Placement via 3D Segmentation Strategy

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Virtual dental implant placement from cone-beam computed tomography (CBCT) is crucial for digital implant surgery.
• Manual placement is complex, requiring balancing restoration orientation, bone adaptation, and anatomical safety.
• Automating this process efficiently presents significant challenges.

Purpose of the Study:

• To develop an intelligent virtual dental implant placement system using a 3D segmentation strategy.
• To automate the generation of virtual implants from CBCT data for missing mandibular first molars.

Main Methods:

• A segmentation module based on nnU-Net was developed to generate virtual implants from CBCT scans.
• An approximation module was employed for mathematical optimization to meet clinical requirements.
• The system was trained and tested on 190 CBCT scans from four centers.

Main Results:

• The AI tool achieved high segmentation accuracy with a surface Dice coefficient (sDice) of 0.903 (internal) and 0.884 (external).
• Average deviations from ground truth were minimal for implant platform (0.82-0.85 mm), apex (1.44-1.47 mm), and angle (4.9-5.5°).
• The tool demonstrated strong performance in predicting virtual implant dimensions and position.

Conclusions:

• The 3D segmentation-based AI tool shows excellent performance in virtual dental implant placement.
• This technology has significant potential for clinical application in digital implant surgery planning.
• The automated approach addresses the complexity of manual implant placement, improving efficiency and accuracy.