Convolution-based truncation correction for C-Arm CT using scattered radiation

Abstract

Patient dose reduction in C-arm computed tomography by volume-of-interest (VOI) imaging is becoming an interesting topic for many clinical applications. One limitation of VOI imaging that remains is the truncation artifact in the reconstructed 3-D volume. This artifact can either be a cupping effect towards the boundaries of the field-of-view (FOV) or an offset in the Hounsfield values of the reconstructed voxels. A new method for the correction of truncation artifacts in a collimated scan is introduced in this work. Scattered radiation still reaches the detector and is detected outside of the FOV, even if axial or lateral collimation is used. By reading out the complete detector area, we can use the scatter signal to estimate the truncated parts of the object: The scattered radiation outside the FOV is modeled as a convolution with a scatter kernel. This new approach is called scatter correction. The reconstruction results using Scatter convolution are at least as good or better than the results with a state-of-the-art method. Our results show that the use of scattered radiation outside the FOV improves image quality by 1.8%.