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

Fully convolutional neural networks like U-Net have been the state-of-the-artmethods in medical image segmentation. Practically, a network is highlyspecialized and trained separately for each segmentation task. Instead of acollection of multiple models, it is highly desirable to learn a universal datarepresentation for different tasks, ideally a single model with the addition ofa minimal number of parameters steered to each task. Inspired by the recentsuccess of multi-domain learning in image classification, for the first time weexplore a promising universal architecture that handles multiple medicalsegmentation tasks and is extendable for new tasks, regardless of differentorgans and imaging modalities. Our 3D Universal U-Net (3D U$^2$-Net) is builtupon separable convolution, assuming that {\it images from different domainshave domain-specific spatial correlations which can be probed with channel-wiseconvolution while also share cross-channel correlations which can be modeledwith pointwise convolution}. We evaluate the 3D U$^2$-Net on five organsegmentation datasets. Experimental results show that this universal network iscapable of competing with traditional models in terms of segmentation accuracy,while requiring only about $1\%$ of the parameters. Additionally, we observethat the architecture can be easily and effectively adapted to a new domainwithout sacrificing performance in the domains used to learn the sharedparameterization of the universal network. We put the code of 3D U$^2$-Net intopublic domain. \url{https://github.com/huangmozhilv/u2net_torch/}

3D U$^2$-Net: A 3D Universal U-Net for Multi-Domain Medical Image Segmentation