English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Whole slide images (WSIs) pose unique challenges when training deep learningmodels. They are very large which makes it necessary to break each image downinto smaller patches for analysis, image features have to be extracted atmultiple scales in order to capture both detail and context, and extreme classimbalances may exist. Significant progress has been made in the analysis ofthese images, thanks largely due to the availability of public annotateddatasets. We postulate, however, that even if a method scores well on achallenge task, this success may not translate to good performance in a moreclinically relevant workflow. Many datasets consist of image patches which maysuffer from data curation bias; other datasets are only labelled at the wholeslide level and the lack of annotations across an image may mask erroneouslocal predictions so long as the final decision is correct. In this paper, weoutline the differences between patch or slide-level classification versusmethods that need to localize or segment cancer accurately across the wholeslide, and we experimentally verify that best practices differ in both cases.We apply a binary cancer detection network on post neoadjuvant therapy breastcancer WSIs to find the tumor bed outlining the extent of cancer, a task whichrequires sensitivity and precision across the whole slide. We extensively studymultiple design choices and their effects on the outcome, includingarchitectures and augmentations. Furthermore, we propose a negative datasampling strategy, which drastically reduces the false positive rate (7% onslide level) and improves each metric pertinent to our problem, with a 15%reduction in the error of tumor extent.

Overcoming the limitations of patch-based learning to detect cancer in whole slide images