The synthetic images were generated by a diffusion model trained on experimental images. The network architecture is based on recent work (A. Nichol, et al 2021) and the training data was taken from Shareloc (W. Quyang, et al 2022).

​

This microtubule does not exist: Super-resolution microscopy image generation by a diffusion model
Alon Saguy, Tav Nahimov, Maia Lehrman, Onit Alalouf, Yoav Shechtman
bioRxiv 2023.07.06.548004; doi: https://doi.org/10.1101/2023.07.06.548004