MFSNet: A multi focus segmentation network for skin lesion segmentation

作者：

Highlights：

• We propose a deep learning model called MFSNet (Multi-Focus Segmentation Network).

• MFSNet uses differently scaled feature maps for skin lesion segmentation.

• It has Boundary Attention, Reverse Attention and Parallel Partial Decoder modules.

• It uses Res2Net as backbone which is a recently proposed CNN model.

• It outperforms past methods when evaluated on PH2, ISIC2017 and HAM10000 datasets.

摘要

•We propose a deep learning model called MFSNet (Multi-Focus Segmentation Network).•MFSNet uses differently scaled feature maps for skin lesion segmentation.•It has Boundary Attention, Reverse Attention and Parallel Partial Decoder modules.•It uses Res2Net as backbone which is a recently proposed CNN model.•It outperforms past methods when evaluated on PH2, ISIC2017 and HAM10000 datasets.

论文关键词：Lesion Segmentation,Deep Learning,Parallel Partial Decoder,Attention Modules,Skin Melanoma

论文评审过程：Received 17 April 2021, Revised 6 November 2021, Accepted 27 March 2022, Available online 1 April 2022, Version of Record 8 April 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108673