Tooth Decay Prediction and Classification from X-Ray Images using Deep CNN

Abstract

In recent days, due to food habits and excessive intake of chocolates, dental diseases occur more often. Because of this, a prediction model that detects dental disease early is to be developed. The dental image processing helps to make accurate selections of early dental illness. Segmentation of dental X-Ray images for the diagnosis of dental diseases is a major research area. In the proposed methodology, the preprocessing of dental images is done by the adaptive approach which is used to level the brightness and contrast of the dental images. Further, it uses the graph- cut segmentation for segmenting the foreground and background of dental X-Ray images. For the prediction of dental images, Deep CNN is used. The different segmentation techniques like Canny Edge detection, Watershed, Threshold, Active contour, Sobel and Otsu’s threshold method are analyzed and compared. The method achieves a greater accuracy of 98% when the graph cut segmentation is induced with Deep CNN.