LCP-Net: A local context-perception deep neural network for medical image segmentation

作者：

Highlights：

• A novel feature extraction module called LCP is proposed, which is consists of PDC and LCE.

• LCP-Net that combines LCP and U-Net is designed for medical image segmentation.

• An improved version of CELoss, based on certainty and deviation distance is proposed.

• Experiments demonstrate that our network outperforms state-of-the-art methods.

摘要

•A novel feature extraction module called LCP is proposed, which is consists of PDC and LCE.•LCP-Net that combines LCP and U-Net is designed for medical image segmentation.•An improved version of CELoss, based on certainty and deviation distance is proposed.•Experiments demonstrate that our network outperforms state-of-the-art methods.

论文关键词：Medical image segmentation,Local context perception,Dilated convolution,Local context embedding

论文评审过程：Received 24 July 2020, Revised 3 November 2020, Accepted 3 November 2020, Available online 6 November 2020, Version of Record 24 January 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114234