Diagnosis and classification of dental caries with deep learning methods in dental radiography

Rapid developments in artificial intelligence and deep learning have attracted attention in the field of health, as in other fields, especially for the early detection and diagnosis of diseases. This research examines the use of Conventional.Neural. Networks..(CNN) for the categorization of dental caries in radiological images. A dataset of 1554 RGB images was used to train and assess a number of models, including CNN, VGG19, Inception, ResNet, and Support Vector Machine..(SVM), to assess classification performance in terms of F1-score,. Doğruluk,.Kesinlik and Duyarlılık. Inception outperformed the other models with a perfect Duyarlılık of 1.00% and an Doğruluk of 97.83%. SVM came in second with 97.4%. VGG19 and ResNet both performed better than the conventional CNN, with accuracies of 86.79% and 88.32%, respectively. CNN's Doğruluk of 82.03% was the lowest. Resampling approaches helped overcome issues like dataset imbalance, and data augmentation enhanced the generalization of the model. According to the study, deeper architectures like ResNet and VGG19 demand more processing power, whereas Inception and SVM models are most suited for classifying dental caries. The findings indicate that improving model performance requires using data augmentation techniques and preserving dataset balance.