Segmentation of muscle fibre images using Voronoi diagrams and active contour models

The segmentation of a complex scene, such as a sample of muscle fibre types, is notoriously difficult and in most published studies on quantitative muscle fibre analysis it was performed manually. We propose a semi-automatic segmentation method which gives good results, similar to those obtained by manual segmentation, but significantly reduces the time necessary for user interaction. Instead of manually tracing each fibre boundary we build a Voronoi diagram based on the set of points which represent the muscle fibre centroids. The edge of each Voronoi polygon is used as the initial position of a snake which is reshaped by an energy minimization process. The final position and shape of snakes closely resemble actual fiber boundaries.