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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The success and generalisation of deep learning algorithms heavily depend onlearning good feature representations. In medical imaging this entailsrepresenting anatomical information, as well as properties related to thespecific imaging setting. Anatomical information is required to perform furtheranalysis, whereas imaging information is key to disentangle scanner variabilityand potential artefacts. The ability to factorise these would allow fortraining algorithms only on the relevant information according to the task. Todate, such factorisation has not been attempted. In this paper, we propose amethodology of latent space factorisation relying on the cycle-consistencyprinciple. As an example application, we consider cardiac MR segmentation,where we separate information related to the myocardium from other featuresrelated to imaging and surrounding substructures. We demonstrate the proposedmethod's utility in a semi-supervised setting: we use very few labelled imagestogether with many unlabelled images to train a myocardium segmentation neuralnetwork. Specifically, we achieve comparable performance to fully supervisednetworks using a fraction of labelled images in experiments on ACDC and adataset from Edinburgh Imaging Facility QMRI. Code will be made available athttps://github.com/agis85/spatial_factorisation.

Factorised spatial representation learning: application in semi-supervised myocardial segmentation