Medical image segmentation method based on genetic neural network and texture information

摘要

No exact and general way has been found to the medical image segmentation because it refers to a special field. For the medical image in texture features, people can estimate the fractal dimension and draw the energy of LAWS, and then optimize RBF and its structure and parameter by way of image division principle in genetic neural network and Hybrid hierarchy genetic algorithm (HHGA). Take the way of clock betting in selection operator; imitate the genetic change of Biological reproduction in variation operator and use two points crossing in intersection operator, which is respectively in the controlling gene and parameter gene. The results show that this method can distinguish different texture and produce good segmentation result when applied to certain medical images.