Challenges in Dental Imaging – Edge and Texture Analysis in Caries Detection

Abstract

One of the most frequent diseases all over the world is represented by dental caries, hence it is of utmost importance to detect them as early as possible, in order to prevent massive tooth decay. Next to clinical examination, radiographic images are essential in identifying caries, especially for lesions located on the contact surface between the posterior teeth. Incipient caries are difficult to identify, sometimes even for specialists, so segmentation and detection through various image processing techniques may be useful. This paper presents an analysis of edge and texture parameters in X-rays containing caries, based on Sobel and Canny operators, Gabor filters and local binary pattern (LBP) operator. Our study set consisted in 80 X-rays from Bitewing Radiography Caries Detection Challenge 2015. Final results align with literature results and confirm the fact that caries segmentation in X-rays is a difficult task, with an average of less than 35% edge pixels correctly identified and poor results for texture segmentation; they also motivate us to refine our work in search of new algorithms and methods for caries segmentation and detection.