An object-independent measure for improving misalignment correction in C-arm CT

Misalignment in c-arm CT is a frequently discussed problem. Numerous methods for misalignment correction were investigated, but up to now no appropriate measure for geometry correction algorithms has been given. We present a new object-in-dependent measure for geometry comparison in c-arm systems. Our approach is based on a virtual object and assigns one single value to compare geometry calibrations. Invariance towards variable calibration phantom positions as well as its independence towards different c-arm systems is demonstrated. We describe the characteristic behaviour for several causes of misalignment and the results of a simulated dataset are presented. (orig.)