Research On Medical Image Segmentation Based On Swarm Intelligence Optimization Algorithms

Recent years, with the development of image technology and computer application technology,computer image technology can be applied more widely in medical field.Now it becomes an important method in clinical diagnosis,and the medical image segmentation is the most important processing step in medical image technology.Therefore,how to segment medical image effectively has currently become major research emphasis home and abroad.This paper proposes an improved medical image segmentation technology, employing hybrid swarm intelligence optimization algorithm to segment normal tissues and morbid tissues in medical image.Hybrid swarm intelligence optimization algorithm uses group strategy to do searching work,paralleling for multi-point at the same time,which makes the searching area more wider and improves global searching ability for groups. Hybrid swarm intelligence optimization algorithm combines artificial fish school algorithm with particle swarm optimization,solving the problem that the artificial fish will converge slowly in the latter phrase and particle swarm will easily fall into local optimum.It applies hybrid algorithm to brain medical image segmentation,using Otsu as its objective function,so segmentation threshold can be found fast and correctly to segment the pathological area.The main work is done as follows:1) We systematically introduce present research situation and the corresponding issues in medical image field home and abroad, and summarize the basic characteristics of medical images and basic methods of medical image segmentation.2) We introduce basic artificial fish and particle swarm algorithm,and analyze their features.For the problem that AFSA converges slowly in the latter phrase,we propose to add a method of variable step into iteration and conduct the experiment to verify feasibility of our approach.3) We propose a hybrid optimal algorithm to segment brain medical image, called IAFS-PSO algorithm,which merges AFSA and PSO together,chooses ostu as the fitness function,optimizes hybrid algorithm to obtain threshold,then segments medical images.Through segment a single threshold,dual-threshold and third threshold for brain medical images,our simulation experiment demonstrates that hybrid algorithm can capture the optimal threshold fast and achieve better image segmentation effect.