Image Reconstruction of Multiple Basis Materials With Data Augmentation in DECT

The objective of the work is to develop algorithms for accurate and stable reconstruction of multiple basis images and virtual monochromatic images (VMIs) in dual-energy CT (DECT).

The reconstruction task remains challenging in DECT imaging cases with basis regions containing up to 3 basis materials, because the number of voxel values, i.e., unknowns, within the basis regions can be considerably larger than the number of measurements. We exploit the volume-conservation (VC) constraint to augment data and develop an optimization-based algorithm for accurate and stable reconstruction of multiple basis images and VMIs in the cases.

Numerical studies are conducted involving simulated and real data for the cases with basis regions containing up to 3 basis materials; and the study result reveals that the algorithm developed can accurately and stably reconstruct multiple basis images and VMIs in DECT.

The exploitation of the VC constraint enables the algorithm proposed for accurately and stably reconstructing multiple basis images in DECT.

The work reveals that it is possible to obtain accurate VMIs and contrast-concentration estimations in DECT with basis regions containing up to 3 basis materials. The algorithm can be extended for obtaining multiple basis images in photon-counting CT and also be applied to dual-energy data collected only over limited-angular range and/or at sparse views.