On-the-fly geometrical calibration fine-tuning of a mobile C-arm CBCT system

C-arm cone-beam computed tomography (CBCT) offers CT-like 3D imaging capabilities while being appropriate for interventional suites. 3D images are reconstructed based on 2D projections gathered during the rotation of the C-arm around the object under investigation. It is mandatory to provide accurate geometrical projection parameters for each acquired 2D image, otherwise significant CT artifacts may be induced in the reconstructed 3D volume. Usually, a pre-calibration (offline calibration) using an X-ray phantom is preformed under the assumption that the C-arm motion is reproducible. Thereby, stochastic misalignments due to the open design of mobile C-arm CBCT systems are not considered. In this article we introduce a novel online calibration algorithm to compensate stochastic mechanical inaccuracies. The performance of the proposed method is demonstrated on simulated and real data.