Home

Awesome

Synthetic Medical Images from Dual Generative Adversarial Networks

Code is split into two stages: a segmentation-mask-generating DCGAN, and an image-to-image translator using pix2pix.

Paper: https://arxiv.org/abs/1709.01872

SynthMed (Repository for GAN-produced synthetic medical images): https://synthmed.github.io/

Pipeline

Flowchart

Prerequisites

Acknowledgements

Stage-I GAN based on: https://github.com/carpedm20/DCGAN-tensorflow <br></br> Stage-II GAN based on: https://github.com/ray0809/pix2pix

Authors: John Guibas, Tejpal Virdi, Peter Li