Improved Artificial Fish Swarm Algorithm And Its Application In Stroke Detection System

Stroke is one of the three major diseases in China.Each year,about 1.3 million people die from stroke in China,tremendously affecting people's qualities of life and health status.Medically,the prevention of stroke is very important,and timely detection of early stroke plays a very important role on the prevention of stroke,which can reduce the damages caused by stroke and increase the probability of curing in the follow-up treatment.Microwave detection is served as a low cost,safe and lightweight detection method.Because the dielectric properties of the human brain tissue are different,especially changing the dielectric distribution caused by the stroke of the brain,the scattering signal of microwave signal resulted from the stroke of the human brain is different.According to the received microwave scattering signal,we can diagnose the stroke by reconstructing the image of the stroke distribution.Based on microwave imaging technology of detecting stroke,we can divided the algorithm procedure into the forward calculation and reverse reconstruction,based on the Finite Difference Time Domain method and swarm optimization algorithm to reconstruct the image of brain dielectric distribution respectively.Owing to the complexity of the internal structure of the brain,the existing microwave imaging technology of detecting stroke is not accurate and the process of calculation is complex.In view of the aforementioned problems,this paper improves the existing Artificial Fish SwarmAlgorithm,which overcomes shortcomings of other algorithms and improves the accuracy of global search and the convergent speed.First of all,this paper analyzes the theoretical basis of microwave detection,the physical structure of the brain and the dielectric properties of each organization.Secondly,this paper investigates the Artificial Fish Swarm Algorithm and puts forward some improvements,including the improvement of adaptive field of vision,the improvement of prey behavior,the improvement of swarm behavior,the improvement of follow behavior and the introduction of random initialized concept.Finally,the efficiency of improved Artificial Fish Swarm Algorithm is proved by the test function.Thirdly,this paper designed the detection system including microwave signal source,microwave transmitting and receiving antenna and stroke detected algorithm based on AFSA.The signal source is selected as the cosine modulation of the ultra wide-band microwave signal.By using the ultra wide-band microwave antenna,we can detect the stroke based on the improved AFSA and reconstruct the image of brain dielectric constant distribution,and then identify and locate the stroke.Fourthly,this paper uses the vector network analyzer and the simplified configuration of the brain model to build the microwave-detection-platform.Then,this paper studies the matching degree of the microwave S-parameter signal of the simulation system and the experimental system based on the correction of the vector network analyzer and the processing of the S-parameter signal.In order to obtain the permittivity of the liquid human brain model based on glycerol,this paper estimates the dielectric constant of the liquid by using the Artificial Fish Swarm Algorithm.Finally,this paper uses the simulated system to verify the stroke-detection-system based on Artificial Fish Swarm Algorithm.The result shows that the algorithm can detect the location of blood clots and the size of that.In this paper,we further compare these effects of AFSA algorithm parameters such as the number of artificial fish and the value of the fitness function of the food density function based on the signal matching algorithm,the influence of the iterative number and the accuracy.Then,we add a template library to improve the detected efficiency of stroke detection that is more than 90% on the basis of existing system.Furthermore,by comparing these applications of Particle Swarm Optimization and Artificial Fish Swarm Algorithm of stroke detection system,we prove that the Artificial Fish Swarm Algorithm has obvious advantages indetecting efficiency,and the detected time is 37% less than that of Particle Swarm Optimization.Finally,we summarize the paper and make prospects.