An Insight Toolkit (ITK) processing framework for segmentation using active contours without edges
is presented in this paper. Our algorithm is based on the work of Chan and Vese [1] that uses level-
sets to accomplish region segmentation in images with poor or no gradient information. The basic idea
is to partion the image into two piecewise constant intensity regions. This work is in contrast to the
level-set methods currently available in ITK which necessarily require gradient information. Similar to
those methods, the methods presented in this paper are also made efficient using a sparse implementation
strategy that solves the contour evolution PDE at the level-set boundary. The framework consists of 6
new ITK filters that inherit in succession from itk::SegmentationFilter. We include 2D/3D example
code, parameter settings and show the results generated on a 2D cardiac image.