Fully automatic segmentation of breast ultrasound images based on breast characteristics in space and frequency domains

作者：

Highlights：

• We propose a reference point (RP) generation algorithm based on breast anatomy.

• We propose a new segmentation framework modeling tumor properties in two domains.

• The proposed frequency constraint is invariant to the brightness and contrast.

• The proposed cost function is graph-representable and can be globally optimized.

摘要

•We propose a reference point (RP) generation algorithm based on breast anatomy.•We propose a new segmentation framework modeling tumor properties in two domains.•The proposed frequency constraint is invariant to the brightness and contrast.•The proposed cost function is graph-representable and can be globally optimized.

论文关键词：Breast ultrasound (BUS) images,Computer-aided diagnosis (CAD),Region of interest (ROI) generation,Automatic segmentation

论文评审过程：Received 18 February 2014, Revised 25 May 2014, Accepted 12 July 2014, Available online 12 August 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.07.026