Weakly supervised learning and interpretability for endometrial whole slide image diagnosis

Mahnaz Mohammadi*, Jessica Cooper, Oggie Arandelovic, Christina Mary Fell, David Morrison, Sheeba Syed, Prakash Konanahalli, Sarah Bell, Gareth Bryson, David James Harrison, David Cameron Christopher Harris-Birtill

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Downloads (Pure)

Abstract

Fully supervised learning for whole slide image based diagnostic tasks in histopathology is problematic due to the requirement for costly and time-consuming manual annotation by experts. Weakly supervised learning which utilises only slide-level labels during training is becoming more widespread as it relieves this burden, but has not yet been applied to endometrial whole slide images, in iSyntax format. In this work we apply a weakly supervised learning algorithm to a real-world dataset of this type for the first time, with over 85% validation accuracy and over 87% test accuracy. We then employ interpretability methods including attention heatmapping, feature visualisation, and a novel end-to-end saliency-mapping approach to identify distinct morphologies learned by the model and build an understanding of its behaviour. These interpretability methods, alongside consultation with expert pathologists, allow us to make comparisons between machine-learned knowledge and consensus in the field. This work contributes to the state of the art by demonstrating a robust practical application of weakly supervised learning on a real-world digital pathology dataset and shows the importance of fine-grained interpretability to support understanding and evaluation of model performance in this high-stakes use case.
Original languageEnglish
Pages (from-to)2025 - 2037
Number of pages13
JournalExperimental Biology and Medicine
Volume247
Issue number22
Early online date25 Oct 2022
DOIs
Publication statusPublished - 1 Nov 2022

Keywords

  • Digital pathology
  • Weak supervision
  • Adenocarcinoma
  • Hyperplasia
  • Endometrial cancer
  • Cancer detection
  • iCAIRD
  • Interpretability
  • XAI

Fingerprint

Dive into the research topics of 'Weakly supervised learning and interpretability for endometrial whole slide image diagnosis'. Together they form a unique fingerprint.

Cite this