A method for improving cone beam computed tomography images using a deep convolution neural network

摘要

Problem to be solved: to provide a system and method for training DCNN to reduce one or more artifacts using a projection space approach or an image space approach.The projection space approach uses artifact contaminated CBCT projection spatial images and corresponding artifact reduced CBCT projection spatial images from each patient in the patient group The DCNN may be trained to include reducing one or more artifacts in the projection space image.The image space approach uses CBCT anatomical images of multiple patients and corresponding registered CT anatomical images from patients' groups and uses multiple CBCT anatomical images and corresponding artifacts reduced CT anatomical images. It can include training DCNN and removing artifacts from CBCT anatomical images.Diagram