Panoramic Dental X-Ray Image Segmentation and Feature Extraction

Panoramic X-Ray images are very popular as a first tool for diagnosis in odontological protocols. Automating the process of analysis of such images is important in order to help dentist procedures. In this process, teeth segmentation from the radiographic images and feature extraction are essential steps. In this paper, we propose a segmentation approach based on mathematical morphology, quadtree decomposition for mask generation, thresholding, snake models. The feature extraction stage is steered by a shape model based on Principal Component Analysis (PCA). Firstly, we take quadtree decomposition after applying low pass filter to smooth the original image and select the smallest blocks to generate a mask. Then, the original image is processed by Otsu’s thresholding. The result is improved by morphological operators and the quadtree mask is applied to address overlapping, a common problem in x-ray images. The obtained regions are searched and the larger ones are selected to find out the teeth candidates. The boundary of the obtained regions are extracted and aligned with the shape model in order to recognize the teeth. The selected curves are used in a search method to initialize a snake technique. Finally, morphometrics data extraction is performed to obtain the teeth measurements for dentist diagnosis. Preliminary experiments show the significance of the proposed method to extract teeth from an image.