Iterative reconstruction for circular cone-beam CT with an offset flat-panel detector

Circular cone-beam computed tomography (CBCT) with a tangentially offset flat-panel X-ray detector offers a large CT field-of-view (FoV) with a relatively small detector. It is used in practice, e.g., for target imaging in image-guided radiotherapy or for localization and attenuation correction in SPECT/CT imaging. The X-ray projections, acquired on a circular source trajectory, each cover roughly half the CT FoV; a central overlap region is imaged by all projections. Offset-detector CBCT reconstruction requires special algorithms. For large detector offsets, previously proposed filtered-backprojection methods can lead to shading artifacts, specifically left/right intensity imbalance. Here, we propose using iterative reconstruction for offset-detector CBCT. To handle the special acquisition geometry, known iterative reconstruction algorithms are modified in terms of axial truncation compensation, redundancy weighting, and algorithm initialization. An efficient implementation using a graphics processing unit (GPU) delivers clinically feasible reconstruction times. Results from patient and phantom studies are presented, showing a clear reduction of artifacts and improvement in image quality.