Inventors at RPI have invented a technology that can overcome the issues seen in few-view CT scans. CT scans use a large number of X-rays exposing the patient to ionizing radiation during this procedure. Though the risk of cancer from a CT scan might be extremely low, there is still a concern that the combined risks of scanning for diagnoses and/or treatment could lead to a greater number of cancers in the future. Few-view CT image reconstruction is an important approach to reduce the ionizing radiation dose. The invention uses a 3D deep-learning-based method for few-view CT image reconstruction from 3D projection data. The proposed method aims at extracting 3D spatial information from 3D projection data, reconstructed from 3D image slices. The invention could additionally reduce the memory requirement for reconstructing an image volume directly from cone-beam projection data by compressing the 3D input into a latent space in a data-driven fashion. The technology then reconstructs the image in a compressed latent space, thereby significantly reducing the computational cost.