| Summary: | 碩士 === 國立成功大學 === 資訊工程學系 === 102 === In recent years, there are more people taking the orthodontic treatment. The treatment process and consuming time are usually different for patients with dissimilar symptoms. For better treatment outcomes, dentists design a set of parameters for orthodontic evaluation. In this thesis, an automated system is proposed to assist dentists in obtaining reliable measurement of these parameters. Dental panoramic radiograph has been widely used in the clinical diagnosis of stomatology. It is used to observe both the teeth and bones in oral cavity. Because the panoramic radiograph provides a wide range of oral view, the detailed information of a single tooth is not always available so that the segmentation of a tooth contour is not so simple.
The proposed system can be divided into four stages: image preprocessing, model training, tooth contour segmentation and assessment of parameters. In the image preprocessing, the image size is first made the same based on the region of oral cavity, the histograms are then normalized, and the regions of interest (ROIs) are locally enhanced at last. There are three terms in the model training stage: the mean shape and mean image model training, the energy function training and the mean weights training. The method to automatically segment the tooth contours by using the statistical model is explained in the stage of tooth segmentation. Finally, the automatic assessment of orthodontic parameters is explained at the last stage.
The experimental results comprise the qualitative and quantitative analyses. The experimental results of the proposed method have been compared to the ground truth from experts for both the tooth contours and the orthodontic parameters. It has been shown that the proposed method can obtain similar accuracy and consistent measurement in helping dentists for treatment evaluation.
|
|---|
