Cleansing Cranio-Facial CT images

Abstract

Objective: When X-ray CT of the craniofacial region with high X-ray absorption rate materials such as metal prosthetics is taken, noise called metal artifacts may appear. To reconstruct a 3-dimensional solid model of bone, segmentation is done by setting a threshold for the CT value. The metal artifact is also extracted with this method. Using image processing software, the artifacts are manually removed from each image, which is complicated and time-consuming. We attempted to remove metal artifacts automatically from craniofacial CT images. Methods: DICOM images of CT archives of patients with head and neck tumors were used. Artifacts and beds were removed from the images, and binarized to obtain bony PNG images. A U-net was trained with the training datasets consist of DICOM images and corresponding PNG images. Results: Metal artifacts were cleansed with high accuracy.