On the Critical Point of Gradient Vector Flow Snake

Segmentation of the left ventricle in cardiac cine MRI using a shape-constrained snake model

“…6 shows the original edge map, the GVC field and the segmentation result, respectively. Since the myocardium is thin and the initial contour should avoid critical points in the force field [44], it is feasible to segment the epicardium using the original force field. Moreover, the original force field flows into the myocardium at the weak epicardium boundaries (see the white ellipse on the upper row in Fig.…”

Snake Validation: A PCA-Based Outlier Detection Method

“…When all snakes have converged to their local minima, we form a pattern image for each snake and project it into the PCA space and compute reprojection error based on only the first principal component. We have compared our PCA-based outlier detection with two automatic initialization techniques: center of divergence [2] and critical point [8] and two gradient-based techniques: GICOV [1] and average directional image gradient computed on evolved snake contours.…”

“…CoD is defined as the point from which the GVF vectors of all the neighboring pixels radiate. Wang et al [8] suggested the idea of critical point (CP) [8], which is defined as the point from which the quantized GVF vectors of its 8-neighborhood do not point to it. Both of these methods tend to initialize more than the desired number of active contours, i.e., the image is over segmented.…”

“…Literature survey shows that much of the effort to date has been exerted on the first two steps, while the last step is practically ignored and left for the application to decide. The automated initialization techniques proposed to date mostly attempt to exploit the structure/shape of the objects (e.g., see [2] and [8]). In this paper we argue that a very crucial step for overall full automation of the snake is necessarily the validation step-often more important than the initialization step-when simple blind/ random initializations are possible.…”

“…We have carried out experiments with two automatic initialization techniques ( [2] and [8]) and simple blind seed initialization. Results show that these automatic initialization techniques produce many undesired seed points.…”