Tubular Surfaces Extraction with Minimal Action Surfaces

This paper presents a new fast approach for surfaces extraction of tubular structures, such as blood vessels, tracheas, and colons. Since they are usually long and thin, when being extracted with the traditional Minimal Action Surface method, edges leakage problem could not be avoided. So a front freezing criterion based on the front's average energy is developed. To freeze the tail of the front whose energy is less than the average energy, and to continue propagating the head of the front whose energy is larger than the average energy. While the head front gets the end of the structure, the whole tubular structure surfaces are extracted. Experimental results show the freezing criterion based on the front's average energy avoids the edges leakage problem effectively. The criterion is based on the global information of the front, so can enhance the robustness to image noise and the local minimum.