ISSMF: Integrated semantic and spatial information of multi-level features for automatic segmentation in prenatal ultrasound images

作者：

Highlights：

• A novel deep learning method is proposed in this paper for segmentation of multiple parts of fetal ultrasound image.

• Multi-level feature fusion module is introduced to the decoder to provide sufficient information for upsampling.

• A novel up sampler (CARAFE) is utilized to fully explore the semantic and spatial information of multi-level features.

• Group normalization is adopted to alleviate the segmentation accuracy degradation caused by the small batch size.

摘要

•A novel deep learning method is proposed in this paper for segmentation of multiple parts of fetal ultrasound image.•Multi-level feature fusion module is introduced to the decoder to provide sufficient information for upsampling.•A novel up sampler (CARAFE) is utilized to fully explore the semantic and spatial information of multi-level features.•Group normalization is adopted to alleviate the segmentation accuracy degradation caused by the small batch size.

论文关键词：Prenatal diagnosis,Ultrasound imaging,Automatic segmentation,Multi-level feature fusion,DeepLabv3+

论文评审过程：Received 23 August 2020, Revised 27 December 2021, Accepted 5 February 2022, Available online 15 February 2022, Version of Record 16 February 2022.

论文官网地址：https://doi.org/10.1016/j.artmed.2022.102254