This paper explores the use of a neural network designed for video-frame interpolation to enhance the resolution of tomographic images. The method, applied to FIB-SEM images of graphene nanosheets, MRI brain scans, and X-ray CT abdominal scans, achieves cubic-voxel resolution and improves the accuracy of 3D tomographic maps. The approach is shown to be versatile and effective across different length scales, offering a valuable image-augmentation strategy for optimizing 3D tomography acquisition.

Laura Gambini, Cian Gabbett, Luke Doolan, Lewys Jones, Jonathan N. Coleman, Paddy Gilligan, Stefano Sanvito