Coupling edge and region-based information for boundary finding in biomedical imagery

We propose in this paper a boundary finding scheme for biomedical imagery which integrates a region-based method and an edge-based technique. We show that more accurate and robust results may be obtained through seeking a joint solution to the traditional approach of curve evolution. The approach incorporates an energy model based on prior distribution and likelihood into the curve evolution of the geodesic active contour (GAC) method. During curve evolution, we use a decision function to adjust relevant parameters in the model automatically so that the curve can easily avoid 'clutter'. For termination of curve evolution, a stability index is proposed which examines curve evolution convergence to ensure that the curve arrives at the boundary robustly and accurately. The experimental results demonstrate that advantages can be achieved using our approach compared to several classical methods.