MIRAU-Net: An improved neural network based on U-Net for gliomas segmentation

作者：

Highlights：

• A novel approach is proposed for brain tumor segmentation enhancement.

• The proposed network combines Residual-Inception modules and Attention module with U-Net architecture.

• The encoder and decoder are connected via a sequence of Inception-Res paths.

• Apply the proposed network to Brats 2019.

摘要

•A novel approach is proposed for brain tumor segmentation enhancement.•The proposed network combines Residual-Inception modules and Attention module with U-Net architecture.•The encoder and decoder are connected via a sequence of Inception-Res paths.•Apply the proposed network to Brats 2019.

论文关键词：Brain tumor segmentation,U-Net,Full convolutional network,Inception,Residual Module,Attention Gate

论文评审过程：Received 24 December 2020, Revised 25 August 2021, Accepted 31 October 2021, Available online 17 November 2021, Version of Record 20 November 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116553