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

Annotating nuclei in microscopy images for the training of neural networks isa laborious task that requires expert knowledge and suffers from inter- andintra-rater variability, especially in fluorescence microscopy. Generativenetworks such as CycleGAN can inverse the process and generate syntheticmicroscopy images for a given mask, thereby building a synthetic dataset.However, past works report content inconsistencies between the mask andgenerated image, partially due to CycleGAN minimizing its loss by hidingshortcut information for the image reconstruction in high frequencies ratherthan encoding the desired image content and learning the target task. In thiswork, we propose to remove the hidden shortcut information, calledsteganography, from generated images by employing a low pass filtering based onthe DCT. We show that this increases coherence between generated images andcycled masks and evaluate synthetic datasets on a downstream nucleisegmentation task. Here we achieve an improvement of 5.4 percentage points inthe F1-score compared to a vanilla CycleGAN. Integrating advancedregularization techniques into the CycleGAN architecture may help mitigatesteganography-related issues and produce more accurate synthetic datasets fornuclei segmentation.

Focus on Content not Noise: Improving Image Generation for Nuclei Segmentation by Suppressing Steganography in CycleGAN