Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Mind the Gap: Scanner-induced domain shifts pose challenges for representation learning in histopathology

Frauke Wilm, Marco Fragoso, Christof A. Bertram, Nikolas Stathonikos, Mathias Öttl, Jingna Qiu, Robert Klopfleisch, Andreas Maier, Marc Aubreville, Katharina Breininger

TL;DR

In insights into the influence of scanner characteristics for downstream applications, the Barlow Triplets are presented to learn scanner-invariant representations from a multi-scanner dataset with local image correspondences and it is shown that self-supervised pre-training successfully aligned different scanner representations.

Abstract

Computer-aided systems in histopathology are often challenged by various sources of domain shift that impact the performance of these algorithms considerably. We investigated the potential of using self-supervised pre-training to overcome scanner-induced domain shifts for the downstream task of tumor segmentation. For this, we present the Barlow Triplets to learn scanner-invariant representations from a multi-scanner dataset with local image correspondences. We show that self-supervised pre-training successfully aligned different scanner representations, which, interestingly only results in a limited benefit for our downstream task. We thereby provide insights into the influence of scanner characteristics for downstream applications and contribute to a better understanding of why established self-supervised methods have not yet shown the same success on histopathology data as they have for natural images.

Mind the Gap: Scanner-induced domain shifts pose challenges for representation learning in histopathology

TL;DR

In insights into the influence of scanner characteristics for downstream applications, the Barlow Triplets are presented to learn scanner-invariant representations from a multi-scanner dataset with local image correspondences and it is shown that self-supervised pre-training successfully aligned different scanner representations.

Abstract

Computer-aided systems in histopathology are often challenged by various sources of domain shift that impact the performance of these algorithms considerably. We investigated the potential of using self-supervised pre-training to overcome scanner-induced domain shifts for the downstream task of tumor segmentation. For this, we present the Barlow Triplets to learn scanner-invariant representations from a multi-scanner dataset with local image correspondences. We show that self-supervised pre-training successfully aligned different scanner representations, which, interestingly only results in a limited benefit for our downstream task. We thereby provide insights into the influence of scanner characteristics for downstream applications and contribute to a better understanding of why established self-supervised methods have not yet shown the same success on histopathology data as they have for natural images.
Paper Structure (8 sections, 2 equations, 4 figures, 1 table)

This paper contains 8 sections, 2 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Materials
  3. Methods
  4. Experiments and Results
  5. Discussion
  6. Conclusion
  7. Compliance with ethical standards
  8. Acknowledgments

Figures (4)

  • Figure 1: Exemplary patch of the multi-scanner dataset.
  • Figure 2: Barlow Triplets. Figure adapted from zbontar2021.
  • Figure 3: Cosine distance of CS2 scanner to seen (NZ 210, NZ 2.0) and unseen (P 1000, GT 450) scanners ($\mu \pm \sigma$ of three repetitions).
  • Figure 4: Concordance of segmentation outputs to the CS2 prediction measured as miou of masks.