Prior wavelet knowledge for multi-modal medical image segmentation using a lightweight neural network with attention guided features

作者：

Highlights：

• Developed a lightweight model to segment various anatomies from multiple modalities.

• The proposed network used the discrete wavelet transform to extract salient features.

• We formulate a new edge-based loss function called FEPE by holding End-Point-Error.

• We proposed an attention-based mechanism to enhance the feature discriminability.

• Compared the segmentation results with eleven state-of-the-art deep learning methods.

摘要

•Developed a lightweight model to segment various anatomies from multiple modalities.•The proposed network used the discrete wavelet transform to extract salient features.•We formulate a new edge-based loss function called FEPE by holding End-Point-Error.•We proposed an attention-based mechanism to enhance the feature discriminability.•Compared the segmentation results with eleven state-of-the-art deep learning methods.

论文关键词：Medical image,Deep learning,Residual shuffle attention,Ultrasound image segmentation

论文评审过程：Received 25 January 2022, Revised 21 May 2022, Accepted 10 July 2022, Available online 16 July 2022, Version of Record 9 August 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118166