Decision Support System for Black Classification of Dental Images Using GIST Descriptors

One of the well-known pathology in the world is dental caries. Dental caries is also called as dental cavities. The prior detection of dental caries helps in decreasing the dental disease rate. Patient care has been improved due to medical image mining. In this paper, abnormal dental images have been classified into various classes based on Black's classification. Graphics and intelligence-based script technology (GIST) descriptor has been used to extract significant information from the dental images. Feature reduction is done using marginal Fisher analysis and then Wilcoxon signed-rank test is used as feature selection method. The classification techniques such as decision tree, fuzzy Sugeno, probabilistic neural network, K-nearest neighbor, AdaBoost and naive Bayes are used for classifying the major and reduced features. According to the results, AdaBoost classifier can best diagnose infected tooth using Black's classification with the classification accuracy of 90, 92% sensitivity and 90% specificity.