Age Estimation based on Dental Radiographs using Hybrid CNN

Abstract

Nowadays, computer vision had been inserted on a large scale in multiple applications in several fields, such as age estimation using dental images. Age estimation based on Convolutional-Neural-Networks (CNN) is a crucial aspect of forensic dentistry, by analyzing the growth and development of teeth. Recently, different techniques based on CNN had been integrated with multiple approaches to extracting a various feature of human and used for age estimation. In this paper, a Convolutional Neural Network was developed according to two scenarios to estimate an age based on dental radiographs. In the first scenario, a CNN classifier is proposed for age estimation from dental images, in the second scenario, the CNN classifier is modified by replacing the last dense layer with a logistic regression classifier to enhance performance. The results showed that the effectiveness of the proposed method in improving the accuracy of age estimation, such that its improved to 95.15 % and the performance measures improved such that: the Precision by 1.83%, the Recall by 2.82% and the F1-score by 2.16%.