Combining various projection access schemes with the algebraic reconstruction technique for low-contrast detection in computed tomography.

We separately applied three different projection access orders (the multilevel scheme (MLS), the sequence access scheme (SAS) and the random permutation scheme (RPS)) to the algebraic reconstruction technique (ART) in an attempt to improve low-contrast object detection in computed tomography (CT). We used both simulated (from various numbers of projections of low-contrast and contrast detail phantoms) and actual data (from megavoltage portal CT) during our study. When coupled with ART, SAS and RPS led to poor low-contrast detection and required multiple iterations for convergence. In contrast, one-iteration MLS yielded the most uniform reconstructions with the highest contrast-to-noise ratios. Therefore, MLS-ART provides the highest dose efficiency of all current reconstruction techniques for imaging low-contrast objects from a small number of projections.