Gender Classification in Panoramic Dental X-Rays using few-shot learning and Ensemble models

Abstract

Integrating advanced technologies has revolutionized dental practice, transforming patient diagnosis and treatment. Radiographic examinations remain one of the most essential tools for identifying hidden dental conditions with precision. According to the American Dental Association (ADA), nearly 90% of dental procedures rely on radiographic analysis, which uncovers hidden issues in over 80% of cases. This study focuses on leveraging Machine learning methodology to improve the diagnostic process by enabling automated classification X-ray images with high accuracy, significantly reducing the time required for manual interpretation. Gender determination from panoramic X-rays plays a crucial role in forensic odontology and clinical diagnosis, as certain diseases are more prevalent in one gender than the other. Studies have shown that statistically significant gender differences exist in craniofacial parameters such as the gonial angle, ramus height, and bigonial width, with these parameters increasing with age. Moreover, this difference is particularly significant on the right side for gonial angle and ramus height [23]. Incorporating gender classification into automated diagnostic systems can further enhance the precision of dental assessments and support more personalized treatment strategies. Our approach integrates various advanced techniques and algorithms, including Few-Shot Learning, Ensemble Models, Convolutional Neural Networks (CNNs), and Transfer Learning models, to get the best model for this dataset. The proposed framework achieved a high accuracy of 96.9%, demonstrating its ability to improve diagnostic precision and efficiency in dental radiography.