Partial Scan Electron Microscopy with Deep Learning

05/31/2019

∙

by   Jeffrey M. Ede, et al.

∙

0

∙

share

We present a multi-scale conditional generative adversarial network that completes 512×512 electron micrographs from partial scans. This allows electron beam exposure and scan time to be reduced by 20× with a 2.6 intensity error. Our network is trained end-to-end on partial scans created from a new dataset of 16227 scanning transmission electron micrographs. High performance is achieved with adaptive learning rate clipping of outlier losses and an auxiliary trainer network. Source code and links to our new dataset and trained network have been made publicly available at https://github.com/Jeffrey-Ede/partial-STEM