Despite advancements in deep learning and computer vision, medical image segmentation is still a challenging problem. A major challenge for many segmentation models is the inherent complexity and inter-connectivity of pixels in medical images. These characteristics require modeling not only local features but also a global understanding of image semantics. In this paper, we propose a deep convolutional neural network called GCEENet to effectively address the above challenges. GCEENet features a combination of global context encoders and local distribution modules, working in conjunction to preserve the global image context. Our experiments on several medical image segmentation datasets show that GCEENet outperforms current state-of-the-art models in all measured metrics.
CITATION STYLE
Hung, N. T., Lan, P. N., Oanh, N. T., Thuy, N. T., & Sang, D. V. (2022). GCEENet: A Global Context Enhancement and Exploitation for Medical Image Segmentation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 13599 LNCS, pp. 141–152). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-20716-7_11
Mendeley helps you to discover research relevant for your work.