Saliency-guided automatic detection and segmentation of tumor in breast ultrasound images. (July 2020)

• Record Type: Journal Article
• Authors: Ramadan, Hiba; Lachqar, Chaymae; Tairi, Hamid
• Is Part Of: Biomedical signal processing and control. Volume 60(2020)
• Publication Date: 2020-07
• Subjects: Computer aided diagnosis -- Breast ultrasound -- Tumor detection -- Tumor segmentation -- Saliency detection -- Graph-based segmentation -- Saliency score -- Active contours; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.101945
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: New hybrid method for automatic detection and segmentation of tumor in breast ultrasound (BUS) images. Saliency is used to emulate radiologist's ability to detect the tumor region and background region in BUS image. Graph-based approach is applied to segment lesion. New saliency score measure is proposed for result refinement. Performance of the proposed method in terms of: tumor detection, tumor segmentation and speed. Abstract: The development of computer aided diagnosis (CAD) systems for the task of tumor detection and segmentation in breast ultrasound (BUS) images presents one of the most active research fields. In this paper, a saliency-guided approach for fast and automatic tumor segmentation in BUS images is proposed. We explore the ability of the concept of saliency to emulate radiologist expertise for tumor lesion detection in BUS images. For that, we compute BUS image saliency to generate effective tumor seeds and background seeds. Then, a graph-based interactive image segmentation method is applied automatically using the generated seeds to extract tumor region. For more accurate segmentation, we propose a novel saliency score measure to apply a post-processing step for result refinement. Our results have been compared with other approaches in challenged datasets and demonstrated the effectiveness of our saliency-guided approach for tumor detection and segmentation using different evaluation metrics.