An Embedded Recurrent Neural Network-based Model for Endoscopic Semantic Segmentation Mahmood Haithami a , Amr Ahmed a , Iman Yi Liao a and Hamid Jalab a a Computer Science Department,University of Nottingham Malaysia Campus b Computer System and Technology Department, University of Malaya Abstract Detecting cancers at their early stage would decrease mortality rate. For instance, detecting all polyps during colonoscopy would increase the chances of a better prognoses. However, endoscopists are fac- ing difculties due to the heavy workload of analyzing endoscopic images. Hence, assisting endoscopist while screening would decrease polyp miss rate. In this study, we propose a new deep learning segmen- tation model to segment polyps found in endoscopic images extracted during Colonoscopy screening. The propose model modifes SegNet architecture to embed Gated recurrent units (GRU) units within the convolution layers to collect contextual information. Therefore, both global and local information are extracted and propagated through the entire layers. This has led to better segmentation performance compared to that of using state of the art SegNet. Four experiments were conducted and the proposed model achieved a better intersection over union łIoUž by 1.36%, 1.71%, and 1.47% on validation sets and 0.24% on a test set, compared to the state of the art SegNet. Keywords SegNet, GRU, Embedded RNN, Polyp Segmentation 1. Introduction According to National Institute of Diabetes and Digestive and Kidney Diseases [1], 60 to 70 million people are afected by a gastrointestinal disease. Malignancies such as esophageal and colorectal cancer are in an increasing rate in western countries [2, 3, 4]. Early detection and removal of such malignant tissues using endoscopy would reduce the risk of developing a cancer. However, endoscopists are facing difculties due to the heavy workload of analyzing endoscopic images [5], subtle lesions, or lack of experience [6]. Therefore, researchers have been proposing deep learning models to help endoscopists marking malignancies during the screening [7]. Polyp segmentation is considered to be a challenging task due to the non-uniformity of the gastrointestinal tract. Hence, researchers tend to modify on-the-shelf deep learning models that proved to be efcient in a specifc domain [8, 9]. Also, the lack of big and representative datasets in the endoscopy domain is a persistent challenge which clamp the performance of 3rd International Workshop and Challenge on Computer Vision in Endoscopy (EndoCV2021) in conjunction with the 18th IEEE International Symposium on Biomedical Imaging ISBI2021, April 13th, 2021, Nice, France  hcxmh1@nottingham.edu.my (M. Haithami); Amr.Ahmed@nottingham.edu.my (A. Ahmed); Iman.Liao@nottingham.edu.my (I. Y. Liao); hamidjalab@um.edu.my (H. Jalab)  0000-0001-6340-8183 (M. Haithami); 0000-0002-7749-7911 (A. Ahmed); 0000-0001-5165-4539 (I. Y. Liao); 0000-0002-4823-6851 (H. Jalab) © 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR Workshop Proceedings http://ceur-ws.org ISSN 1613-0073 CEUR Workshop Proceedings (CEUR-WS.org)