A hybrid edge-based segmentation approach for ultrasound medical images. (January 2017)

• Record Type: Journal Article
• Authors: Gupta, Deep; Anand, R.S.
• Is Part Of: Biomedical signal processing and control. Volume 31(2017)
• Page Start: 116
• Page End: 126
• Publication Date: 2017-01
• Subjects: Ultrasound -- Distance regularized level set -- Edge based active contour -- Kernel fuzzy C-means -- Segmentation; 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.2016.06.012
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: This paper presents a hybrid edge-based segmentation method for ultrasound medical images. The proposed method is based on the kernel fuzzy clustering and active contour model using distance regularized level set function. The proposed method helps to remove the need of manual intervention and also increase the averaged computational time. Abstract: Ultrasound imaging is one of the most widely used and the cheapest diagnostic tools of medical imaging modalities. In this paper, a hybrid approach for accurate segmentation of the ultrasound medical images is presented that utilizes both the features of kernel fuzzy clustering with spatial constraints and edge based active contour method using distance regularized level set (DRLS) function. The result obtained from the kernel fuzzy clustering is utilized not only to initialize the curve that spreads to identify the estimated region or object boundaries, but also helps to estimate the optimal parameters, which are responsible for controlling the level set evolution. The DRLS formulation also increase the processing speed by removing the need of re-initialization of the level set function. The performance of the proposed method is evaluated by conducting the several experiments on both the synthetic and real ultrasound images. Experimental results show that the proposed method improves the segmentation accuracy and also produces better results by successfully segmenting the object boundaries compared to others.