Tooth segmentation using Gaussian mixture model and genetic algorithm

Background: The present study suggested an image segmentation method for dental cone beam computed tomography (CBCT) data with a proposed preprocessing step and genetic algorithm. Segmentation of dental CT images is often hampered by the proximity of teeth and alveolar bones that display similar brightness. The present study sought to overcome this difficulty by using a Gaussian mixture model (GMM) and contrastlimited adaptive histogram equalization (CLAHE) in the preprocessing step. First, the original dental image was processed by GMM to eliminate regions other than the teeth and alveolar bones. Then, we composed the preprocessed image by enhancing tooth contours through application of CLAHE. Finally, tooth and pulp regions were extracted via the evolutionary process of genetic algorithm. We confirmed that tooth segmentation using a genetic algorithm was effective in segmenting teeth that are adjacent and have similar shapes and brightness.