Detection of low-dose CT reconstruction artifacts using a bi-modal approach

Low-dose Computed Tomography (CT) has the benefit of exposing patients to less radiation. However, low dose CT requires special reconstruction techniques to improve the clarity of the image. Unfortunately, these special reconstruction techniques often cannot remove all of the low-dose artifacts. It is important to recognize these artifacts else we run the risk of obscuring important detail or adding false features. In this work, we present a simple scheme which allows us to detect these artifacts. Our technique applies to the specific low-dose CT strategy in which the number of X-ray views taken from the patient is reduced. The first step uses directional interpolation in the low dose sinogram to add more views. While the image created from this interpolated sinogram does not have any artifacts it lacks significantly in clarity due to blurring. Our scheme then compares this image with the image created directly with a low-dose CT reconstruction technique which has better detail but also some remaining artifacts. The comparison reveals these artifacts which we then remove by simple pixel replacement.