Research on Segmentation Algorithm for 3D Medical Image Based on Neural Network

This paper proposed a new segmentation algorithm of 3D medical image based on Hopfield neural networks. The method makes Hopfield neural networks and genetic algorithm (GA) combined and use texture analysis to achieve segmentation of tumor CT image. This method not only overcome shortcoming that the Hopfield neural network easily converge to local optimal solution, but also break the constraints that use gray of lesion site and external information, improve the quality of the segmentation algorithm and robustness. Through an experiment justified the method is effective and it realizes computer-aided diagnosis system.