Multi-material Reconstruction Method Based On Deep Prior of Spectral Computed Tomography

Abstract

Spectral computed tomography (Spectral CT) has attracted more and more attention because of its ability of material discrimination. However, as the number of materials increases, it becomes more difficult to decompose the material according to the polychromatic projection. This paper presents a direct multi-material reconstruction method, in which a deep convolutional neural network (CNN)-based prior is incorporated into the optimization model. The efficient iterative algorithm is designed under the framework of the alternating direction method of multipliers (ADMM). The numerical experiments further validate the superiority of the proposed method in multi-material reconstruction and noise suppression.