Contour extraction in medical images using initial boundary pixel selection and segmental contour following

In today's health care, an imaging system plays an important role throughout the entire clinical process from diagnosis and treatment planning to surgical procedures and follow-up studies of disease. Boundary detection is a technique used to segment an object within a region of interest in the medical image for further clinical applications. Contour extraction is one of the most important boundary detection methods. In this paper, an object contour extraction for gray-level medical images using automatic initial boundary pixel selection and tracing of a segmental contour based on the boundary pixels obtained by the initial boundary pixel selection is proposed. Experimental results on artificial images of convex and deep concave objects, and real CT and MRI images show that, in comparing with other existing methods, a more detailed and accurate contour can be obtained using the proposed object contour extraction method. This has low computational complexity, which will benefit applications to clinical diagnosis, treatment, surgery, and follow up studies.