Segmentation of teeth in CT volumetric dataset by panoramic projection and variational level set

Abstract

Purpose Quantification of teeth is of clinical importance for various computer assisted procedures such as dental implant, orthodontic planning, face, jaw and cosmetic surgeries. In this regard, segmentation is a major step. Methods In this paper, we propose a method for segmentation of teeth in volumetric computed tomography (CT) data using panoramic re-sampling of the dataset in the coronal view and variational level set. The proposed method consists of five steps as follows: first, we extract a mask in a CT images using Otsu thresholding. Second, the teeth are segmented from other bony tissues by utilizing anatomical knowledge of teeth in the jaws. Third, the proposed method is followed by estimating the arc of the upper and lower jaws and panoramic re-sampling of the dataset. Separation of upper and lower jaws and initial segmentation of M. Hosntalab (B) · A. Abbaspour Tehrani-Fard teeth are performed by employing the horizontal and vertical projections of the panoramic dataset, respectively. Based the above mentioned procedures an initial mask for each tooth is obtained. Finally, we utilize the initial mask of teeth and apply a Variational level set to refine initial teeth boundaries to final contours. Results The proposed algorithm was evaluated in the presence of 30 multi-slice CT datasets including 3,600 images. Experimental results reveal the effectiveness of the proposed method. Conclusion In the proposed algorithm, the variational level set technique was utilized to trace the contour of the teeth. In view of the fact that, this technique is based on the characteristic of the overall region of the teeth image, it is possible to extract a very smooth and accurate tooth contour using this technique. In the presence of the available datasets, the proposed technique was successful in teeth segmentation compared to previous techniques.

A Hybrid Segmentation Framework for Computer-Assisted Dental Procedures

“…Another parameter which is used for evaluation of the proposed algorithm is the receiver operating characteristic (ROC) curve [49], which is more accurate than the two conventional segmentation techniques including thresholding, watershed, and two previous works [23], [25]. This is illustrated in Fig.…”

“…In the previous work [23], we introduced a multi-step method based on the level set for teeth segmentation of CT images. In addition, we proposed a new technique based on panoramic re-sampling and variational level set for teeth segmentation in CT volumetric data in [24], [25]. In this research, inspired by our previous experiences, considering the imaging constraint of dental CT and anatomical knowledge of teeth and jaws, we propose a hybrid technique for teeth segmentation in multi-slice computed tomography (MSCT) dataset.…”

“…We perform the panoramic image of the dataset by coronal projection of the re-sampled data [25]. In the available datasets, due to symmetry of upper and lower jaws, the whole upper and lower teeth are visible in the panoramic images.…”

“…In previous works [24], [25], the segmented teeth were visualized by MEDAL software [45]. In this research, a surface rendering algorithm is applied to the classified slices for the purpose of volumetric visualization.…”

Dental segmentation in cone-beam computed tomography images using watershed and morphology operators

“…This technique was effective to acquire an accurate and smooth teeth contour. [3] Momeni and Aghaeizadeh Zoroofi introduced an automated method based on wavelet descriptors to dental recognition in multislice computed tomography (MSCT) data. Experimental results reveal that their proposed technique is effective to automatically teeth classification.…”