Detection and Classification of Peri-Implant Marginal Bone Loss in Cone-Beam Computed Tomography Using a Deep...

Area of Science:

• Radiology
• Artificial Intelligence
• Dental Implantology

Background:

• Peri-implant marginal bone loss is a key factor in dental implant failure.
• Radiographic assessment of bone loss can be challenging and requires expert interpretation.
• Cone-beam computed tomography (CBCT) offers detailed 3D imaging but generates large datasets.

Purpose of the Study:

• To evaluate a YOLOv8 deep learning model for automated detection and grading of peri-implant marginal bone loss.
• To assess the model's performance on 2D images derived from CBCT scans.

Main Methods:

• A retrospective study utilized 699 2D CBCT sections.
• A YOLOv8 object-detection model was trained to identify implants and bone loss.
• Performance was evaluated using accuracy, precision, recall, and F1-score metrics.

Main Results:

• The YOLOv8 model demonstrated strong diagnostic performance with an overall accuracy of 0.90.
• High performance was observed for healthy sites (F1 0.95) and mild lesions (F1 0.90).
• The model achieved a mean average precision (mAP@0.5) of 0.889 and excellent reliability (Kappa = 0.954) for a three-class scheme.

Conclusions:

• A YOLOv8-based deep learning model reliably detects and grades peri-implant marginal bone loss on CBCT images.
• The model shows potential for automating the analysis of dental implant radiographs.
• Further research should focus on expanding datasets and validating performance in diverse clinical settings.