diff --git a/README.md b/README.md index a2275da..dd9da97 100644 --- a/README.md +++ b/README.md @@ -15,7 +15,7 @@ The results of this work illustrate the potential of GAN-based methods to synthe Code, data and other resources to reproduce this work are available at \url{https://github.com/budai4medtech/midl2023}. ![fig](short-paper/figures/main-results/outputs/drawing-v00.png) -Results from Diffusion-Super-resolution-GAN (DSR-GAN) and transformer-based-GAN (TB-GAN): +**Figure** Results from Diffusion-Super-resolution-GAN (DSR-GAN) and transformer-based-GAN (TB-GAN): (a) Training losses for Generator and Discriminator networks, (b) FID scores, and (c) 256x256 pixel size trans-cerebellum images of two randomised batches (B1, B2) of real and synthesised (DSR-GAN and TB-GAN).