Synthesized Data Generation for Enhanced Machine Learning in Dental X-ray Analysis: A Novel Approach to Age and Gender Prediction

M. I. R. Shuvo, Awal Ahmed Fime, Samuel Kim, Hannah Lee, Ian Jung, Jong-Hoon Kim, Young-Jin Jung, Dong-Hun Han

Abstract: In the field of dental radiography, the precise estimation of age and gender based on X-ray pictures holds significant importance for a range of applications. Nevertheless, the scarcity of reliable and robust data presents a significant obstacle. This research highlights the significance of both data quality and data quantity in the context of deep learning models. The conventional methods of data augmentation are insufficient when used for dental X-ray pictures. To tackle this issue, we propose the introduction of an innovative data synthesis technique, which involves the extraction of dental structures from X-ray images and subsequently generates synthesized visual representations. Satisfactory results for age and gender prediction are achieved by employing the VGG19 and ResNet50 models, utilizing the synthesized data generated within the framework of the proposed research. The present study highlights the importance of comprehensive datasets and proposes a novel approach to improve the precision of dental radiography predictions.

External IDs:dblp:conf/icbea2/ShuvoFKLJKJH24