Towards annotation-efficient segmentation via image-to-image translation

Eugene Vorontsov, Pavlo Molchanov, Matej Gazda, Christopher Beckham, Jan Kautz, Samuel Kadoury

Abstract: Highlights•A new semi-supervised training method with a weak supervision component is proposed.•Proposed semi-supervised method tested with different encoder–decoder architectures.•Segmentation decoder reused for generation, can be trained in absence of annotations.•Modified segmentation tasks (brain, liver, synthetic) with ‘healthy’ vs ‘diseased’ cases.