Automatic superpixel-based segmentation method for breast ultrasound images
作者：
Highlights：
• A two-phase, superpixel-based method is proposed for segmenting BUS images.
• The first phase employs coarse superpixels to achieve effective tumor detection.
• The second phase uses fine superpixels to enable accurate tumor segmentation.
• Both phases employ edge- and region-based information to segment the superpixels.
• The results show that the method enables automatic and accurate tumor segmentation.
摘要
•A two-phase, superpixel-based method is proposed for segmenting BUS images.•The first phase employs coarse superpixels to achieve effective tumor detection.•The second phase uses fine superpixels to enable accurate tumor segmentation.•Both phases employ edge- and region-based information to segment the superpixels.•The results show that the method enables automatic and accurate tumor segmentation.
论文关键词：Ultrasound breast images,Superpixels,Graph cuts segmentation,Active contour models,Support vector machine,Breast cancer diagnosis
论文评审过程：Received 28 May 2018, Revised 27 October 2018, Accepted 13 November 2018, Available online 22 November 2018, Version of Record 15 December 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.11.024
