Transfer Learning-Based Automated Tooth and Anomaly Classification in OPG X-ray Images

Recent advancements in computer science and machine learning have revolutionised the landscape of Medical Science, particularly in leveraging algorithms to support healthcare professionals. This study introduces a novel method for automatically detecting and classifying teeth and anomalies from Orthopantomogram (OPG) X-ray images utilising transfer learning techniques. Our approach focused on distinguishing between two types of teeth (incisors and molars) and identifying two anomalies (cap and impacted teeth). By harnessing the capabilities of pre-trained deep learning models, specifically VGG16 and InceptionV3, we developed specialised models tailored to this task. Through fine-tuning a dedicated dataset of OPG X-ray images, we achieved classification accuracies of 98% and 99% for VGG16 and InceptionV3, respectively. These findings underscore the potential of these models to significantly enhance dental diagnostics by offering precise and efficient automated tooth classification. Such advancements promise to streamline diagnostic workflows, mitigate human error, and ultimately improve accessibility to reliable dental care solutions. Future research will expand this work by enlarging the dataset and exploring advanced architectures to further enhance the classification accuracy and robustness.