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

Unmasking the decision-making process of machine learning models is essentialfor implementing diagnostic support systems in clinical practice. Here, wedemonstrate that adversarially trained models can significantly enhance theusability of pathology detection as compared to their standard counterparts. Welet six experienced radiologists rate the interpretability of saliency maps indatasets of X-rays, computed tomography, and magnetic resonance imaging scans.Significant improvements were found for our adversarial models, which could befurther improved by the application of dual batch normalization. Contrary toprevious research on adversarially trained models, we found that the accuracyof such models was equal to standard models when sufficiently large datasetsand dual batch norm training were used. To ensure transferability, weadditionally validated our results on an external test set of 22,433 X-rays.These findings elucidate that different paths for adversarial and real imagesare needed during training to achieve state of the art results with superiorclinical interpretability.

Advancing diagnostic performance and clinical usability of neural networks via adversarial training and dual batch normalization