Impact of image constraints on DECT reconstruction with limited-angular-range data

There exists interest in DECT with limited-angular- range (LAR) data for meeting unique workflow needs in practical imaging applications. Optimization-based, one-step algorithm can lead to quantitatively accurate image reconstruction in DECT with advanced LAR scanning configurations where data are col- lected with non-overlapping LAR scans. The aim of this work is to investigate and evaluate the impact of different image constraint designs in the optimization program on the reconstructed images qualitatively and quantitatively. We investigate and prototype optimization programs that incorporate two different designs of image constraints on the directional-total variation (DTV) of either basis images or monochromatic images. We then efficiently derive algorithms for numerically solving these optimization programs, thanks to the primal-dual-based algorithm framework. Results from both simulated and real data collected using a two-orthogonal-arc scanning configuration suggest that one-step algorithms may be designed that can reconstruct images with minimal artifacts from LAR data. Furthermore, the image constraint on the DTVs of monochromatic images may yield images that are visually closer to the reference images from FAR data, while the constraint on the DTVs of basis images may yield improved accuracy of iodine concentration estimation. The methods and results in the work may engender insights into the design of practical reconstruction procedures for DECT with LAR data of potential application significance.