A Random Forest-based Automatic Classification of Dental Types and Pathologies using Panoramic Radiography Images

DENTAL TYPES AND ANOMALIES CLASSIFICATION USING RANDOM FOREST

Abstract

Detection and classification of tooth types and anomalies is crucial for accurate dental assessment, treatment planning, and overall oral health preservation. The integration of machine learning expedites the classification of tooth types and anomalies with lesser manual intervention, streamlining the diagnostic process and enhancing overall efficiency in dental healthcare. This study addresses the automated classification of oral types and anomalies in panoramic radiograph images which is a challenging task due to the complexity and variability of oral conditions. We propose an innovative approach using hyperparameter-optimized Random Forest ensemble learning achieved through Genetic Algorithm. Various pre-processing techniques are applied to enhance data quality by removing null values and minimizing noise in image data followed by feature extraction. The model consists of 10 Decision Trees trained on various subsets, with hyperparameters systematically optimized using a Genetic Algorithm. Outcomes from individual Decision Trees are aggregated through majority voting. Comprehensive experimental assessment, including an 80-5-15 training-validation-testing split, results in an impressive 98% accuracy. This highlights the effectiveness and superiority of our approach and demonstrates the potential of Random Forest ensemble learning for automated accurate classification, with the added benefit of Genetic Algorithm-driven hyperparameter tuning for improved model performance. Our approach may be applied in real-time for telemedicine and can help dental professionals make decisions about prevention, diagnosis, and treatment planning.