Fully convolutional attention network for biomedical image segmentation
作者：
Highlights：
• We propose a fully convolutional attention network (FCANet) that enhances the feature representation of biomedical images by aggregating context information from long-range and short-range distances.
• Lightweight space and channel attention modules are proposed. These modules can be embedded in any end-to-end network to improve the segmentation effect.
• The segmentation effect of FCANet on three open datasets is improved, including the Chest X-ray collection, Kaggle 2018 data science bowl and Herlev dataset.
摘要
•We propose a fully convolutional attention network (FCANet) that enhances the feature representation of biomedical images by aggregating context information from long-range and short-range distances.•Lightweight space and channel attention modules are proposed. These modules can be embedded in any end-to-end network to improve the segmentation effect.•The segmentation effect of FCANet on three open datasets is improved, including the Chest X-ray collection, Kaggle 2018 data science bowl and Herlev dataset.
论文关键词：Biomedical image,Segmentation,Dilated fully convolutional network,Attention modules,Long-range and short-range distance
论文评审过程：Received 5 February 2020, Revised 3 June 2020, Accepted 3 June 2020, Available online 5 June 2020, Version of Record 8 June 2020.
论文官网地址：https://doi.org/10.1016/j.artmed.2020.101899
