Machine learning for classification of periodontal defects (vertical versus horizontal) using CBCT datasets.

Area of Science:

• Dentistry
• Radiology
• Artificial Intelligence

Background:

• Accurate classification of periodontal bone defects is crucial for effective treatment planning.
• Conventional radiographic methods face limitations in complex anatomical areas.

Purpose of the Study:

• To develop and validate machine learning models for automatic classification of periodontal bone defects.
• To assess the performance of different machine learning algorithms using CBCT data.

Main Methods:

• Trained convolutional neural networks, random forest, support vector machines, and gradient boosting classifiers.
• Utilized radiomic features and raw image data from 1,847 teeth in 312 patients.
• Evaluated performance using five-fold cross-validation against expert consensus.

Main Results:

• The convolutional neural network achieved the highest accuracy (91.4%).
• Achieved 89.8% sensitivity for vertical defects and 92.6% specificity for horizontal defects.
• Demonstrated an area under the ROC curve of 0.946.

Conclusions:

• Machine learning, especially deep learning, reliably classifies periodontal defect morphology on CBCT images.
• These models can enhance diagnostic consistency and clinical decision-making in periodontology.