Keywords: digital pathology, foreground segmentation, semi-supervised learning, GANs
TL;DR: A semi-supervised training approach to train cross-stain foreground segmentation models in digital pathology from single-stain annotation
Abstract: The analysis of whole computational pathology slides can often be accelerated by excluding background areas from the analysis. Deep learning has proven to be superior to signal processing techniques to robustly recover the foreground in HE Images. However, naively generalizing this technique to the wide variability of histological stains used in practice would require annotations in all stain domains. To avoid this, we propose a method which leverages tissue annotation from a single stain to perform foreground segmentation in slides with other non-annotated stains.
Registration: I acknowledge that acceptance of this work at MIDL requires at least one of the authors to register and present the work during the conference.
Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
Paper Type: novel methodological ideas without extensive validation
Primary Subject Area: Segmentation
Secondary Subject Area: Transfer Learning and Domain Adaptation
Confidentiality And Author Instructions: I read the call for papers and author instructions. I acknowledge that exceeding the page limit and/or altering the latex template can result in desk rejection.