Predicting root fracture after root canal treatment and crown installation using deep learning.

Area of Science:

• Dentistry
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Vertical root fracture (VRF) is a common cause of tooth loss after root canal treatment and crown placement.
• Predicting VRF is complex due to its multifactorial nature.
• Accurate prediction of VRF is crucial for preserving teeth and improving treatment outcomes.

Purpose of the Study:

• To develop and evaluate a deep learning (DL) model for predicting the likelihood of vertical root fracture (VRF).
• To compare the performance of a deep neural network (DNN) model against a support vector machine (SVM) model in VRF prediction.
• To identify key parameters influencing VRF for improved predictive accuracy.

Main Methods:

• A dataset of 145 clinical cases (97 fractured, 48 non-fractured teeth) from a five-year period was analyzed.
• Deep learning techniques, specifically a DNN model, were employed to analyze 17 mixed-type tabular features.
• The DNN model's predictive performance was compared with a support vector machine (SVM) model.

Main Results:

• The DNN model achieved a higher accuracy (80.7%) and F1-score (0.857) compared to the SVM model (71.7% accuracy, 0.817 F1-score).
• The DNN model utilizing 17 features demonstrated superior performance over models with fewer features.
• The study identified significant characteristics contributing to root fracture prediction using the DNN model.

Conclusions:

• Deep learning models show significant potential for predicting vertical root fracture (VRF) post-root canal therapy and prosthesis.
• The developed DL model can assist clinicians in assessing fracture risk more effectively.
• Implementing DL in clinical practice may lead to improved decision-making and better patient outcomes.