Scaling calibration in the ATRACT algorithm

Abstract

Recently, a reconstruction algorithm for region of interest (ROI) imaging in C-arm CT was published, named Approximated Truncation Robust Algorithm for Computed Tomography (ATRACT). Even in presence of severe data truncation, it is able to reconstruct images without the use of any explicit extrapolation or prior knowledge. However, this method suffers from a scaling artifact in the reconstruction. In this paper, we have investigated a calibration applied in the projection domain to compensate this scaling problem. The proposed correction method is evaluated by using six clinical datasets in presence of different artificial truncation. The results shows that a relative root mean square error (rRMSE) of up to 0.9% is achieved by the corrected ATRACT method.