Bottleneck feature supervised U-Net for pixel-wise liver and tumor segmentation

作者：

Highlights：

• Propose a U-Net with dense module, inception module and dilated convolution.

• Propose a bottleneck feature supervised U-Net.

• Merge the attention map into the loss function to focus more on liver border.

摘要

•Propose a U-Net with dense module, inception module and dilated convolution.•Propose a bottleneck feature supervised U-Net.•Merge the attention map into the loss function to focus more on liver border.

论文关键词：CNN,Liver tumor,Segmentation,U-Net,Encoding,Bottleneck

论文评审过程：Received 18 September 2019, Revised 28 November 2019, Accepted 10 December 2019, Available online 11 December 2019, Version of Record 18 December 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.113131