| Summary: | This paper presents a classification of malocclusion using convolutional neural networks (CNN) and knowledge-based systems (KBS). Malocclusion is a dental abnormality and occlusal feature that deviates from the ideal occlusion. Early detection and treatment can resolve the malocclusion problem. However, the process of detection takes time, and there are also cases of misdiagnosis of the malocclusion in referral letters for an orthodontist. Thus, the use of machine learning (ML) algorithms is needed to facilitate the detection and derive a more accurate classification of the malocclusion. The dataset has undergone pre-processing that includes standardising the image size and several processes of augmentation. The dataset used for this research is 1,064 intraoral images of 236 patients. The system was developed using the ResNet-50 model with an accuracy of 61.20%, precision, recall, and F1-score values of 68%, 61%, and 60%, respectively. This research has developed a system that is able to classify malocclusion based on the IOTN that is currently being used by orthodontists. The classification emphasises classifying the five main grades of IOTN, which include grades Very Great, Great, Moderate, Little, and None. In conclusion, all three objectives of this research have been accomplished, and each objective's deliverables are detailed in this research. In the future, a better system for classifying malocclusion based on the IOTN should be developed. © 2023 IEEE.
|
|---|