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

Unsupervised anomaly detection (UAD) methods are trained with normal (orhealthy) images only, but during testing, they are able to classify normal andabnormal (or disease) images. UAD is an important medical image analysis (MIA)method to be applied in disease screening problems because the training setsavailable for those problems usually contain only normal images. However, theexclusive reliance on normal images may result in the learning of ineffectivelow-dimensional image representations that are not sensitive enough to detectand segment unseen abnormal lesions of varying size, appearance, and shape.Pre-training UAD methods with self-supervised learning, based on computervision techniques, can mitigate this challenge, but they are sub-optimalbecause they do not explore domain knowledge for designing the pretext tasks,and their contrastive learning losses do not try to cluster the normal trainingimages, which may result in a sparse distribution of normal images that isineffective for anomaly detection. In this paper, we propose a newself-supervised pre-training method for MIA UAD applications, named PseudoMulti-class Strong Augmentation via Contrastive Learning (PMSACL). PMSACLconsists of a novel optimisation method that contrasts a normal image classfrom multiple pseudo classes of synthesised abnormal images, with each classenforced to form a dense cluster in the feature space. In the experiments, weshow that our PMSACL pre-training improves the accuracy of SOTA UAD methods onmany MIA benchmarks using colonoscopy, fundus screening and Covid-19 ChestX-ray datasets. The code is made publicly available viahttps://github.com/tianyu0207/PMSACL.

Self-supervised Pseudo Multi-class Pre-training for Unsupervised Anomaly Detection and Segmentation in Medical Images