Red-GAN: Attacking class imbalance via conditioned generation. Yet another medical imaging perspective

Ahmad B Qasim, Ivan Ezhov, Suprosanna Shit, Oliver Schoppe, Johannes Paetzold, Anjany Sekuboyina, Florian Kofler, Jana Lipkova, Hongwei Li, Bjoern Menze

Keywords: GAN, class imbalance, data augmentation, segmentation

Abstract:

Exploiting learning algorithms under scarce data regimes is a limitation and a reality of the medical imaging field. In an attempt to mitigate the problem, we propose a data augmentation protocol based on generative adversarial networks. The networks are conditioned at a pixel-level (segmentation mask) and at a global-level information (acquisition environment or lesion type). Such conditioning provides immediate access to the image-label pairs while controlling class specific appearance of the synthesized images. To stimulate synthesis of the features relevant for the segmentation task, an additional passive player in a form of segmentor is introduced into the adversarial game.
We validate the approach on two medical datasets: BraTS, ISIC. By controlling the class distribution through injection of synthetic images into the training set we achieve control over the accuracy levels of the datasets' classes.

Track: full conference paper

Paper Type: both

Presentation Upload Agreement: I agree that my presentation material (videos and slides) will be made public.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/red-gan-attacking-class-imbalance-via/code)