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

Histopathological image segmentation is a laborious and time-intensive task,often requiring analysis from experienced pathologists for accurateexaminations. To reduce this burden, supervised machine-learning approacheshave been adopted using large-scale annotated datasets for histopathologicalimage analysis. However, in several scenarios, the availability of large-scaleannotated data is a bottleneck while training such models. Self-supervisedlearning (SSL) is an alternative paradigm that provides some respite byconstructing models utilizing only the unannotated data which is oftenabundant. The basic idea of SSL is to train a network to perform one or manypseudo or pretext tasks on unannotated data and use it subsequently as thebasis for a variety of downstream tasks. It is seen that the success of SSLdepends critically on the considered pretext task. While there have been manyefforts in designing pretext tasks for classification problems, there haven'tbeen many attempts on SSL for histopathological segmentation. Motivated bythis, we propose an SSL approach for segmenting histopathological images viagenerative diffusion models in this paper. Our method is based on theobservation that diffusion models effectively solve an image-to-imagetranslation task akin to a segmentation task. Hence, we propose generativediffusion as the pretext task for histopathological image segmentation. We alsopropose a multi-loss function-based fine-tuning for the downstream task. Wevalidate our method using several metrics on two publically available datasetsalong with a newly proposed head and neck (HN) cancer dataset containinghematoxylin and eosin (H\&E) stained images along with annotations. Codes willbe made public athttps://github.com/PurmaVishnuVardhanReddy/GenSelfDiff-HIS.git.

GenSelfDiff-HIS: Generative Self-Supervision Using Diffusion for Histopathological Image Segmentation