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

Colorectal cancer, caused by an unusual growth of tissues in a body called polyp, is the third most prevailing cancer worldwide and remained the second most cause of deaths by cancer in 2020. Early stage detection of the cancer can prevent the deaths. Computer Aided Diagnosis (CAD) system could be a major breakthrough for early detection of the cancer. The system uses image processing techniques. Among the image processing techniques segmentation has a great value. The diagnostic process results are highly dependent on the accuracy of performed segmentation. Nowadays, many supervised and unsupervised techniques are used for the task of segmentation. Deep neural networks have outperformed other state-of-the-art approaches for the task. In this paper, we present an end-to-end deep neural network for segmentation of polyps in images. The network is modified version of the U-Net architecture. The network being much more memory efficient than the U-Net architecture, inferences segmentation of the images more accurate than the U-Net. We reduce number of layers of the U-Net architecture both in the encoding and decoding path, and introduce residual blocks and batch normalization in the encoding path to prevent learning of redundant features, to avoid over-fitting and to accelerate the training process, and in the decoding path to avoid gradient vanishing issue in long dependence of the neural network during training we use bi-directional long short term memory network with batch normalization. We train and validate the network on Kvasir dataset for the task. The network accurately segments the polyp part in the images with 92.46% test accuracy.

A Modified Memory-Efficient U-Net for Segmentation of Polyps