Partial volume artifact reduction in computed tomography

In the computed tomography (CT) head scans, dark streaks connecting the petrous bones are often observed. These streaks can lead to severely degraded image quality and render the image unusable for diagnosis. In this paper, we analyze one the main sources of the streaking artifact the nonlinear partial volume. We have developed an algorithm that effectively combats these artifacts. The algorithm estimates the degree of partial voluming by calculating the variations of the bony structures between adjacent slices. It then performs the correction based on a quadratic error model. Various experiments were conducted to demonstrate the effectiveness of the correction scheme