Technical Studies, The Ultrasound Image Segmentation Based On Active Contour Models

Since 1895 Roentgen discovered X beam, the progress of clinical medicine has beenvery fast. Ultrasonic is widely applied in clinical examination because of no invasive, nopains, non-radioactive rays damages, and it is fast, accurate, convenient, characteristics andeconomic. The ultrasonic image segmentation is an important method for clinical medicine,so the ultrasonic image segmentation is a very important topic to study. At present, thesegmentation widely applied in the ultrasonic imaging system is based on the thresholdvalue method and based on doctor the manual division method.This paper will introduce the ultrasonic image segmentation based on Active ContourModel. In general, Active Contour Models include parametric active contour model andgeometric active model. Kass proposed parametric active contour model, and it can detectimage edge by minimize the energy function of spline near the ROI (region of interest).Parametric model has been successfully applied in medical image segmentation, but it iseasy to fall into local minima and can not deal with topological changes.In view of the parameter active contour model insufficiency, the scholars proposed thelevel set method, which has the topology to be the possible change characteristic, hasremedied the insufficiency of parameter active contour model. We mainly carried outgeodesic active contour and Mumford-Shah model. Level set methods based onMumford-Shah model are the excellent and important methods based on deformable model.Because of its reliance on global information of homogeneous regions in the image, it cansegment the image quickly and precisely. But the traditional level set methods must bere-initialized level set functions costly so that level set functions can be retained to aSDF(Signed Distance Function)and image can be effectively segmented. While, in orderto retain the SDF, the time step must be short ,and the evolution procedure wastes verymuch time. This paper through importing a new energy item way to reduce the iterationnumber and the iteration time without the needs of re-initialization Through theexperiment ,we find that the improvement algorithm can greatly reduce the iterationnumber while retain the image segmentation mass.