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

Self-supervised pre-training has become the priory choice to establishreliable neural networks for automated recognition of massive biomedicalmicroscopy images, which are routinely annotation-free, without semantics, andwithout guarantee of quality. Note that this paradigm is still at its infancyand limited by closely related open issues: 1) how to learn robustrepresentations in an unsupervised manner from unlabelled biomedical microscopyimages of low diversity in samples? and 2) how to obtain the most significantrepresentations demanded by a high-quality segmentation? Aiming at theseissues, this study proposes a knowledge-based learning framework (TOWER)towards enhanced recognition of biomedical microscopy images, which works inthree phases by synergizing contrastive learning and generative learningmethods: 1) Sample Space Diversification: Reconstructive proxy tasks have beenenabled to embed a priori knowledge with context highlighted to diversify theexpanded sample space; 2) Enhanced Representation Learning: Informativenoise-contrastive estimation loss regularizes the encoder to enhancerepresentation learning of annotation-free images; 3) Correlated Optimization:Optimization operations in pre-training the encoder and the decoder have beencorrelated via image restoration from proxy tasks, targeting the need forsemantic segmentation. Experiments have been conducted on public datasets ofbiomedical microscopy images against the state-of-the-art counterparts (e.g.,SimCLR and BYOL), and results demonstrate that: TOWER statistically excels inall self-supervised methods, achieving a Dice improvement of 1.38 percentagepoints over SimCLR. TOWER also has potential in multi-modality medical imageanalysis and enables label-efficient semi-supervised learning, e.g., reducingthe annotation cost by up to 99% in pathological classification.

A Knowledge-based Learning Framework for Self-supervised Pre-training Towards Enhanced Recognition of Biomedical Microscopy Images