Multi-scale multi-hierarchy attention convolutional neural network for fetal brain extraction

作者：

Highlights：

• A 3D multi-scale multi-hierarchy attention convolutional neural network (MSMHA-CNN) is developed for fetal brain extraction in MR images.

• A multi-scale feature learning block is proposed to learn the contextual features of highresolution in-plane slice and contextual features between slices of the fetal brain MR images with an-isotropic resolution.

• We evaluate the proposed MSMHA-CNN method on collected fetal brain MR images. Experimental results show the excellent performance of our method.

摘要

•A 3D multi-scale multi-hierarchy attention convolutional neural network (MSMHA-CNN) is developed for fetal brain extraction in MR images.•A multi-scale feature learning block is proposed to learn the contextual features of highresolution in-plane slice and contextual features between slices of the fetal brain MR images with an-isotropic resolution.•We evaluate the proposed MSMHA-CNN method on collected fetal brain MR images. Experimental results show the excellent performance of our method.

论文关键词：Fetal brain extraction,In utero MR images,Multi-scale,Multi-hierarchy,3D convolutional neural network

论文评审过程：Received 24 March 2022, Revised 26 June 2022, Accepted 4 September 2022, Available online 7 September 2022, Version of Record 22 September 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.109029