Guidance for cone-beam CT design: tradeoff between view sampling rate and completeness of scanning trajectories

Abstract

When an imaging task is specified, the design of a cone-beam CT scanner includes specifications of the scanning trajectory and corresponding image reconstruction algorithms, requirements on the detector size, and requirements on the x-ray tubes. Given the limited flat-panel detector readout speed and the need of short scanning time in a clinical setting, the available number of total view angles is normally limited to several hundred. It is known that when all the focal spots are distributed along a circular trajectory, the cone-beam artifacts are present in the reconstructed out-of-plane images when the cone-angle is relatively large. In order to mitigate or eliminate the cone-beam artifacts, the source trajectory should be complete in the sense of satisfying the so-called Tuy data sufficiency condition. However, assuming a constant number of view angles, a complete source trajectory will potentially lead to a lower view sampling rate and cause view aliasing artifacts. Therefore, for a given imaging task and a given total number of view angles, it is important to study the tradeoff between the view sampling rate and the completeness of the scanning source trajectories. In this paper, we numerically and experimentally studied the above tradeoff. Specifically, numerical simulations were conducted to study this tradeoff using three different source trajectories: (1) a circular trajectory, (2) a helical trajectory, and (3) a two-concentric-orthogonal-circle trajectory. A single x-ray tube and a flat panel imager mounted on an optical bench was utilized to experimentally study the tradeoff between the circular source trajectory and the helical source trajectory. For the complete source trajectories, some novel cone-beam image reconstruction algorithms have been utilized to reconstruct images and compare image quality in numerical simulations and benchtop experiments.