Search for a command to run...
One of the most widespread oral diseases in the world is dental caries, and in order to avoid serious complications and loss of teeth, it should be detected in time and correctly. Conventional ways of diagnosis are based on visual inspection and radiographic evaluation by dental care providers and may take a long time and may be prone to inconsistency. To overcome these problems, this paper presents an ensemble Convolutional Neural Network (CNN)-based architecture that is used in the automatic detection of dental caries on the basis of dental radiographs. The offered framework combines several CNN models that together can use their complementary advantages in features extraction and classification, thus increasing detection and robustness. The model makes use of pre-trained CNN networks like ResNet, DenseNet, and EfficientNet with a transfer learning to derive deep features of dental images. These models are optimized using standard dental caries labelled datasets and predictions are clustered with ensemble methods, including weighted averaging and majority voting. Normalization, contrast enhancement, and data augmentation are image preprocessing techniques that are used to enhance generalization of models. The experimental outcomes prove that the ensemble model works better than the single CNN models in accuracy, precision, recall, and F1-score and minimizes false positives and false negatives. The suggested solution is a valid and effective dental caries detector that will identify changes in oral health and assist clinical decision-making based on the automated detection of dental caries.