Dual Neural Network Approach for Dual-Energy Imaging and Material Decomposition
摘要
Dual-energy computed tomography (DECT) is a promising technology which uses X-ray beams with different energies to obtain the material attenuation information of the scanned object. However, in contrast to single-energy CT (SECT), DECT imaging suffers from high hardware costs, complex systems and increased radiation doses. In order to obtain DECT images easily, in this paper, we propose a dual neural network structure that achieves spectral mapping and material decomposition. Specifically, we developed the mapping U-Net and the material decomposition U-Net, respectively. Finally, we verified the feasibility of the proposed method on the clinical cranial cavity. Numerical results show that the proposed method can obtain high-quality DECT images from SECT images and realize accurate basis material decomposition.