Image to Image translation : Pre to post contrast 2D MRI Image synthesis (base model pix2pix) #Deep learning

This project is already been done with the baseline pix2pix model. But the result is not good. As in the generated post contrast images are not very much similar to the real post contrast images. The SSIM between the fake and target image is low. As both the source and target images are pretty similar. I will provide the code and the dataset looking for someone who can improve the quality such making the model only focusing on the contrast etc. As the model is very light it doesnt take more than 2 hours to train.