using a GAN to generate image with mask for image segmentation

I'm working on an image segmentation problem and I didn't have too many images for training.

Is it possible to create a GAN that create image with their respective mask?