Please use this identifier to cite or link to this item:
Title: Rethinking Pooling Operation for Liver and Liver-Tumor Segmentations
Authors: Lei, J
Lei, T
Zhao, W
Xue, M
Du, X
Nandi, AK
Keywords: image segmentation;deep learning;weighted pooling;U-net;skip connection
Issue Date: 10-Jan-2022
Publisher: Frontiers Media SA
Citation: Lei, J., Lei, T., Zhao, W., Xue, M., Du, X. and Nandi, A.K. (2022) 'Rethinking Pooling Operation for Liver and Liver-Tumor Segmentations', Frontiers in Signal Processing, 1, 808050, pp. 1-10. doi: 10.3389/frsip.2021.808050
Abstract: © 2022 Lei, Lei, Zhao, Xue, Du and Nandi. Deep convolutional neural networks (DCNNs) have been widely used in medical image segmentation due to their excellent feature learning ability. In these DCNNs, the pooling operation is usually used for image down-sampling, which can gradually reduce the image resolution and thus expands the receptive field of convolution kernel. Although the pooling operation has the above advantages, it inevitably causes information loss during the down-sampling of the pooling process. This paper proposes an effective weighted pooling operation to address the problem of information loss. First, we set up a pooling window with learnable parameters, and then update these parameters during the training process. Secondly, we use weighted pooling to improve the full-scale skip connection and enhance the multi-scale feature fusion. We evaluated weighted pooling on two public benchmark datasets, the LiTS2017 and the CHAOS. The experimental results show that the proposed weighted pooling operation effectively improve network performance and improve the accuracy of liver and liver-tumor segmentation.
Other Identifiers: 808050
Appears in Collections:Dept of Mechanical Aerospace and Civil Engineering Research Papers

Files in This Item:
File Description SizeFormat 
FullText.pdf1.93 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons