歯科用パノラマX線画像におけるニューラルネットワークを用いた金属領域の分離

Dental panoramic X-ray images are commonly used for the diagnosis of dental diseases. To make a panoramic image, a tomosynthesis method is a powerful tool for imaging the focal plane of interest. However, the panoramic X-ray image basically consists of the integral of the linear attenuation coefficients of several organs along the Xray path, so that the image quality of the front teeth deteriorates due to the shadow of the cervical vertebrae. In order to solve this problem, we proposed a method of removing the cervical vertebrae from a panoramic X-ray image. However, in this process, metallic objects caused large distortion in the final image. In this paper, we proposed a method to identify the metallic objects using an artificial neural network and photon counting data collection