Prostate Segmentation from 2-D Ultrasound Images Using Graph Cuts and Domain Knowledge

In this paper we present a graph cuts based segmentation technique that incorporates the domain knowledge based fuzzy inference system to find the prostate boundary more accurately. By using this prior knowledge, we increase the robustness of the algorithm at weak boundaries which are common in ultrasound images. Also in traditional graph cuts algorithm, corrections on segments will be done by user after the first run, but in the proposed method there is no user interaction after initialization and we use the priors to add hard constraints for the second run of the graph cuts.