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

Although deep learning can provide promising results in medical imageanalysis, the lack of very large annotated datasets confines its fullpotential. Furthermore, limited positive samples also create unbalanceddatasets which limit the true positive rates of trained models. As unbalanceddatasets are mostly unavoidable, it is greatly beneficial if we can extractuseful knowledge from negative samples to improve classification accuracy onlimited positive samples. To this end, we propose a new strategy for buildingmedical image analysis pipelines that target disease detection. We train adiscriminative segmentation model only on normal images to provide a source ofknowledge to be transferred to a disease detection classifier. We show thatusing the feature maps of a trained segmentation network, deviations fromnormal anatomy can be learned by a two-class classification network on anextremely unbalanced training dataset with as little as one positive for 17negative samples. We demonstrate that even though the segmentation network isonly trained on normal cardiac computed tomography images, the resultingfeature maps can be used to detect pericardial effusion and cardiac septaldefects with two-class convolutional classification networks.

Building Disease Detection Algorithms with Very Small Numbers of Positive Samples