Age Group Classification of Dental Radiography without Precise Age Information Using Convolutional Neural Networks

Area of Science:

• Forensic Dentistry
• Artificial Intelligence in Healthcare
• Radiographic Imaging Analysis

Background:

• Accurate age estimation from dental radiographs is crucial for forensic science and personalized oral healthcare.
• Deep neural networks (DNNs) have improved age estimation accuracy but require extensive labeled datasets, which are often unavailable.
• This research addresses the challenge of age estimation when precise age labels are limited.

Purpose of the Study:

• To investigate the efficacy of a deep neural network (DNN) model for estimating tooth age using panoramic dental radiographs without precise age information.
• To evaluate the performance of the DNN model with image augmentation techniques.

Main Methods:

• Development of a DNN model for age estimation from dental radiographs.
• Application of an image augmentation technique to enhance the dataset.
• Classification of 10,023 images into age groups (10s to 70s).
• Validation using 10-fold cross-validation and calculation of accuracies with varying tolerances (±5, ±15, ±25 years).

Main Results:

• The DNN model achieved high accuracies: 53.846% (±5 years), 95.121% (±15 years), and 99.581% (±25 years).
• The probability of the estimation error exceeding one age group was found to be 0.419%.
• The model demonstrated effective age estimation capabilities despite the absence of precise age labels.

Conclusions:

• Artificial intelligence, specifically DNNs, shows significant potential for reliable tooth age estimation from dental radiographs.
• The developed model is applicable in both forensic investigations and clinical oral healthcare settings.
• Image augmentation and DNNs can overcome limitations of small labeled datasets in dental age estimation.