Title: One-step inversion algorithms for spectral CT, with application to dynamic Cone Beam CT

Abstract: Dual Energy Cone Beam Computed Tomography (DE-CBCT) is a promising technique for several medical applications, including dynamic angiography. Recently, a dynamical two-step method has been proposed : first, the water and iodine projections are computed from the multi-energy sinograms, then, a dynamic image of the iodine contrast is reconstructed using 4D Total-Variation (TV) constrained reconstruction from the iodine projections. In contrast to the 2-step methods, one-step methods use a model relating directly the multi-material images to the multi-energy sinograms. This kind of methods are well-known to reduce the noise correlation between the material images by avoiding the intermediate decomposition step, but request to solve an non-convex large scale optimization problem which can be challenging. In this work we use the Non-Linear Primal-Dual Hybrid Gradient Method (NL-PDHGM) optimization framework to propose two versions of a one-step method which is based on an empirical model : the first one is a version which considers the multi-material images as static object whereas the second version is developped for a specific application with a static water image and a dynamic iodine image consisting of a series of 3D iodine images associated to different time points. This last version is developped to obtain the evolution of the iodine concentration in the blood vessels during a single CBCT scan. To evaluate the proposed one-step methods we used simulations which consider a CBCT system with dual layer spectral detector and a brain phantom with a static and a dynamic vascular tree. The proposed one-step methods are compared with 2-step methods.