Exploring AI-driven deep learning approaches for optimizing space detection in single gap implantation based on CBCT...

Area of Science:

• Dental imaging and AI
• Computer-assisted surgery
• Medical image analysis

Background:

• Computer-assisted implant surgery (CAIS) improves accuracy but faces challenges in clinical implementation due to time-consuming manual annotation.
• Existing AI solutions have limited evaluation for single edentulous regions, highlighting a gap in automated pre-surgical planning tools.

Purpose of the Study:

• To introduce a novel, automated three-step approach for detecting single edentulous areas and adjacent structures.
• To leverage AI-driven segmentation without manual intervention for improved efficiency in pre-surgical planning.

Main Methods:

• Utilized nnUNet-based DentalSegmentator for anatomical structure segmentation on 66 CBCT scans (80 edentulous regions).
• Employed morphological image processing, Watershed algorithm, particle analysis, and a decision tree for automated detection.
• Validated performance against expert annotations and analyzed computational time.

Main Results:

• Achieved high performance metrics: accuracy (0.95), precision (0.95), recall (0.96), specificity (0.93), and F1-score (0.96).
• Automated detection time was significantly reduced to 3.3 seconds, a 37-fold improvement over expert annotation (120 seconds).

Conclusions:

• The automated detection approach is reliable, time-efficient, and consistent for identifying edentulous areas.
• This method represents a valuable advancement for pre-surgical planning in computer-assisted implant surgery.