Automatically synthesizing dental panoramic image from dental CBCT via maxilla and mandible separation

Abstract

In dental clinics today, panoramic radiograph and cone-beam computed tomography (CBCT) are often utilized to aid in diagnosis and treatment planning. Compared to CBCT, panoramic radiograph is an effective modality for cases with many metal implants. However, it suffers from superimposition and soft-tissue overlaps. To resolve the innate issues of panoramic radiograph, CBCT volume is used to synthesize panoramic radiograph. Commercial methods of synthesizing panoramic radiograph from CBCT volume requires the user to identify the dental arch and applies the ray-sum method on a cylindrical assumption. The resulting synthesized panoramic images often fail to show all the teeth fully and the images are severely blurred. In this paper, a new method to synthesize a panoramic image automatically from a dental CBCT volume is proposed. The method is comprised of four parts: dental arch creation, orthogonal slice creation, maxilla-mandible separation, panoramic image synthesis. The dental arch was detected automatically by segmenting the teeth information from maximum intensity projection using k-mean clustering and fitting a parabola. From the parabola, orthogonal slices were created. Then, the maxilla and mandible were separated by identifying the dividing points from the orthogonal slices. The maxilla and mandible were fitted independently using least square fitting to generate a panoramic image. The resulting image was successful in capturing all teeth, nerve and jaw information. In addition, since this method did not implement the ray-sum method, the resulting image was not blurred.