DEVELOPMENT OF DEEP LEARNING MODELS FOR DENTAL DISEASE DETECTION AND CLASSIFICATION OF DENTAL DISEASE LEVELS WITH RGB DENTAL IMAGE INPUT

• Main Author: Milandga Milenio, Rizka
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/81235
• Institution: Institut Teknologi Bandung

Teeth are a vital part of the human body yet only only 4.5% of the population has regular dental check-ups. There are two dental diseases that are still the focus and there are still many cases that occur throughout the world, namely caries and periodontitis. Previous dental disease detection studies have used x-ray image data and only a few have used RGB image data, but unfortunately there are only 2 studies that provide publicly accessible datasets. Research on the classification of dental caries disease levels and calculus has never been done before. This research developed a deep learning model to detect dental diseases and classify disease levels using RGB dental imagery. With a new publicly accessible dataset, dental numbering algorithm, and four deep learning models, the study achieved 90.8% precision for numbering dentures, an average mAP value of 62% for dental disease detection, and obtained an average accuracy score of 57% for the classification of dental disease levels. These findings demonstrate the potential of the model in supporting early diagnosis of dental disease, which could facilitate better prevention and treatment, and could pave the way for the development of more accessible and efficient diagnostic tools for dental health.