A computer
system, a non-transitory computer readable medium, and a method for
computed tomography (CT) reconstruction. The computer
system is configured to perform the method and the computer readable medium is configured to store instructions for performing the method. The method includes obtaining two or more spectral CT measurement data of different materials of a target from one or more 2D or 3D spectral CT images acquired by a
CT imaging system using more than one
CT imaging energy regimes; and producing a pair reconstructed spectral CT images that are representative of densities of the different materials of the target by performing one or more iterations on the two or more spectral CT measurement data using a trained
machine learning network that is trained to performing a
diffusion sampling operation and a forward-model-based likelihood on one or more intermediate data sets representing the two or more spectral CT measurement data.