Tumor saliency estimation for breast ultrasound images via breast anatomy modeling

作者：

Highlights：

• A novel optimization framework is proposed for tumor saliency estimation.

• Breast anatomy is modeled to generate accurate foreground and background maps.

• A new objective function in the optimization framework is developed to deal with images without salient objects.

• The newly proposed method achieves the best performance among nine saliency estimation models using two datasets.

摘要

•A novel optimization framework is proposed for tumor saliency estimation.•Breast anatomy is modeled to generate accurate foreground and background maps.•A new objective function in the optimization framework is developed to deal with images without salient objects.•The newly proposed method achieves the best performance among nine saliency estimation models using two datasets.

论文关键词：Tumor saliency estimation,Breast ultrasound (BUS),Breast anatomy modeling

论文评审过程：Received 27 December 2020, Revised 19 June 2021, Accepted 16 August 2021, Available online 20 August 2021, Version of Record 3 September 2021.

论文官网地址：https://doi.org/10.1016/j.artmed.2021.102155