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/

Zendy Plus

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

Zendy Tools

Zendy Open

Convolutional Neural Networks (CNNs) achieve excellent computer-assisteddiagnosis with sufficient annotated training data. However, most medicalimaging datasets are small and fragmented. In this context, GenerativeAdversarial Networks (GANs) can synthesize realistic/diverse additionaltraining images to fill the data lack in the real image distribution;researchers have improved classification by augmenting data with noise-to-image(e.g., random noise samples to diverse pathological images) or image-to-imageGANs (e.g., a benign image to a malignant one). Yet, no research has reportedresults combining noise-to-image and image-to-image GANs for furtherperformance boost. Therefore, to maximize the DA effect with the GANcombinations, we propose a two-step GAN-based DA that generates and refinesbrain Magnetic Resonance (MR) images with/without tumors separately: (i)Progressive Growing of GANs (PGGANs), multi-stage noise-to-image GAN forhigh-resolution MR image generation, first generates realistic/diverse 256 X256 images; (ii) Multimodal UNsupervised Image-to-image Translation (MUNIT)that combines GANs/Variational AutoEncoders or SimGAN that uses a DA-focusedGAN loss, further refines the texture/shape of the PGGAN-generated imagessimilarly to the real ones. We thoroughly investigate CNN-based tumorclassification results, also considering the influence of pre-training onImageNet and discarding weird-looking GAN-generated images. The results showthat, when combined with classic DA, our two-step GAN-based DA cansignificantly outperform the classic DA alone, in tumor detection (i.e.,boosting sensitivity 93.67% to 97.48%) and also in other medical imaging tasks.

Combining Noise-to-Image and Image-to-Image GANs: Brain MR Image Augmentation for Tumor Detection