Neutrosophic Set-Based Caries Lesion Detection Method to Avoid Perception Error

Dental caries is an infectious oral disease. The monitoring of caries region boundary, in regular intervals, is important for treatment purpose. To detect dental caries lesion, most of the time dentists use X-ray images. Due to human brain perception, sometimes it is difficult to detect the caries lesion accurately by observing the X-ray image manually. In this work, a framework has been proposed to detect caries lesion automatically within the optimum time. Almost all caries detection methods from the radiographic images apply iterative methods upon the entire image to separate initially suspected regions. Then, further processing is performed on the separated regions. This method reduced huge computation efforts by avoiding applying iterative methods upon the entire input image. This method transforms the input X-ray image into its equivalent neutrosophic domain to obtain initially suspected region. We have used a custom feature named ‘weight’ for neutrosophication. This feature is calculated by fusing other features differently. Once the initial region is detected, it is examined further to test whether there exists any iso-center rings like the catchment basin. This is the most important property of caries lesions. After the suspected region is confirmed as caries lesion, then the caries boundary is detected using active contour technique. The advantage of this system is that it avoids repetitive iterations at the time of suspected region selection using neutrosophication. Repetitive iterations upon the entire picture dimension are a time-consuming job. The performance of the proposed research work is satisfactory; the average accuracy is above 92%.