Related Experiment Video
Updated: Oct 29, 2025
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
1.1K
Prediction of Radiation-Related Dental Caries Through PyRadiomics Features and Artificial Neural Network on Panoramic
Vanessa De Araujo Faria1, Mehran Azimbagirad2, Gustavo Viani Arruda1
1Department of Radiology, Faculty of Medicine, University of São Paulo, Ribeirao Preto, Sao Paulo, Brazil.
Journal of Digital Imaging
|July 13, 2021
Summary
This study introduces an AI neural network method to predict and detect radiation-related caries in head and neck cancer patients undergoing radiotherapy. The AI model achieved high accuracy, offering a reliable tool to decrease side effects on susceptible teeth.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Radiation-related caries (RRC) are a significant side effect of head and neck cancer (HNC) radiotherapy (RT).
- Effective prediction and detection methods for RRC are currently lacking.
- Managing RRC is crucial for improving patient quality of life during and after HNC treatment.
Purpose of the Study:
- To propose and evaluate an artificial intelligence (AI) based method for predicting and detecting RRC in HNC patients undergoing RT.
- To utilize features extracted from panoramic radiographs for caries analysis.
- To assess the accuracy and reliability of the AI model in identifying and predicting RRC.
Main Methods:
- Retrospective analysis of panoramic dental images from 15 HNC patients (420 teeth).
- Manual labeling of teeth for caries presence (caries/non-caries) and RRC susceptibility (resistant/vulnerable) by two dentists.
- Extraction of 105 statistical/morphological image features using PyRadiomics.
- Application of an artificial neural network (ANN) classifier for feature selection and tooth labeling (caries detection and RRC prediction).
Main Results:
- The AI model demonstrated high accuracy in detecting RRC with 98.8% sensitivity (AUC=0.9869).
- The model achieved 99.2% accuracy in predicting RRC (AUC=0.9886).
- The proposed method provides a reliable tool for early identification and prediction of RRC before RT initiation.
Conclusions:
- The AI-powered method using neural networks and PyRadiomics features offers a reliable approach for predicting and detecting RRC.
- This method can aid in mitigating the adverse effects of RT on susceptible teeth.
- The findings suggest the potential for proactive management of RRC in HNC patients undergoing radiotherapy.
