Accurate segmentation of the temporomandibular joint (TMJ) in dental CBCT images is a necessary process for quantitative analysis of related diseases such as temporomandibular joint osteoarthritis (TMJ-OA). A TMJ segmentation method is proposed using the small field-of-view CBCT images. Combining with the characteristics of gray distribution in image field-of-view, the method adopts binarization, connected component analysis and morphological processing to segment image. Experimental results of 86 cases of left TMJ demonstrate the segmentation accuracy is satisfactory compared with the manual annotation, with a Dice coefficient of 0.96. The proposed method is promising in segmenting the condyle and future study is warranted to verify the reliability of the proposed method.

3D image segmentation,Temporomandibular joint,Cone Beam Computed Tomography