Guided image generation for improved surgical image segmentation

Emanuele Colleoni, Ricardo Sanchez-Matilla, Imanol Luengo, Danail Stoyanov

Abstract: Highlights•Segmentation models for computer vision in surgery requires manual annotations.•Generative adversarial networks can generate synthetic data to support segmentation.•Surgery-GAN is proposed to produce synthetic surgical images from segmentation maps.•Synthetic images improve localization of under-represented surgical structures.