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

Recently, some pioneering works have preferred applying more complex modulesto improve segmentation performances. However, it is not friendly for actualclinical environments due to limited computing resources. To address thischallenge, we propose a light-weight model to achieve competitive performancesfor skin lesion segmentation at the lowest cost of parameters and computationalcomplexity so far. Briefly, we propose four modules: (1) DGA consists ofdilated convolution and gated attention mechanisms to extract global and localfeature information; (2) IEA, which is based on external attention tocharacterize the overall datasets and enhance the connection between samples;(3) CAB is composed of 1D convolution and fully connected layers to perform aglobal and local fusion of multi-stage features to generate attention maps atchannel axis; (4) SAB, which operates on multi-stage features by a shared 2Dconvolution to generate attention maps at spatial axis. We combine four moduleswith our U-shape architecture and obtain a light-weight medical imagesegmentation model dubbed as MALUNet. Compared with UNet, our model improvesthe mIoU and DSC metrics by 2.39% and 1.49%, respectively, with a 44x and 166xreduction in the number of parameters and computational complexity. Inaddition, we conduct comparison experiments on two skin lesion segmentationdatasets (ISIC2017 and ISIC2018). Experimental results show that our modelachieves state-of-the-art in balancing the number of parameters, computationalcomplexity and segmentation performances. Code is available athttps://github.com/JCruan519/MALUNet.

MALUNet: A Multi-Attention and Light-weight UNet for Skin Lesion Segmentation