Composite Feature Set Based Dental Image Segmentation Framework through Unsupervised Learning.

Dental X-Ray (DXR) image segmentation has gained a great research interest due to its significance in various dental related applications such as Dental Diseases Detection, Caries detection, etc. This paper proposes a new DXR image segmentation framework based on Adaptive Fuzzy C-Means Clustering which considers not only the gray-level pixel intensities but also the proximity features of dental image such Edges, Boundaries, Color and Binary features. A new background detection mechanism is also proposed to further boost up the segmentation performance. Simulation experiments conducted over a standard DXR image dataset shows the performance effectiveness. Further, the performance of proposed approach is measured through performance metrics such as Specificity, Sensitivity, False Positive Rate, Positive Predictive Value, False Discovery Rate and False Negative Rae proves the outstand performance. The average detection rate of proposed mechanism is observed as 85.21% and the average FPR is observed as 14.79%. [ABSTRACT FROM AUTHOR]