Application Of Primate Intelligent Optimization Algorithm In Medical Image Segmentation

In recent years,with the development of medical technology,medical imaging plays an increasingly important role in clinical diagnosis,and computer-aided processing of medical images directly affects doctors’ judgment of disease diagnosis and lesion severity.Therefore,image processing technology has become increasingly important in medical diagnosis and treatment,where image segmentation techniques can accurately locate lesion areas,thereby improving the accuracy and efficiency of disease diagnosis.Threshold segmentation techniques,due to their low cost,fast speed,and high efficiency,occupy a crucial position in the field of medical image segmentation.However,traditional image thresholding segmentation techniques face severe challenges in terms of segmentation speed and quality with increasing image quantities.On the other hand,swarm intelligence optimization algorithms can effectively and quickly find optimal solutions for complex multidimensional nonlinear problems.Therefore,this thesis combines traditional image thresholding segmentation techniques with swarm intelligence optimization algorithms to improve the precision and effectiveness of image segmentation.Considering various intelligent optimization algorithms developed in recent years,this thesis focuses on the chimpanzee optimization algorithm and the capuchin search algorithm,from different perspectives to improve them using multiple strategies.The improved chimpanzee optimization algorithm and the capuchin search algorithm are then applied to medical image segmentation to enhance its effects and speed.The main contents of this thesis include:(1)A summary of domestic and foreign research on medical imaging and an overview of mainstream medical image segmentation methods.(2)Based on the chimpanzee optimization algorithm,this thesis proposes two improvement strategies.Firstly,we propose the Modified Chimp Optimization Algorithm(MCh OA),which integrates the Logistic chaos sequence and simplex method to optimize the population initialization method and improve the quality of the population by upgrading the worst individuals in the population.Secondly,we propose the Improved Chimp Optimization Algorithm(ICh OA),which integrates the Sin chaos sequence and Cauchy mutation strategy to optimize the population initialization method and improve the global search capability of the algorithm.The position updating strategy is also improved using the Cauchy mutation to balance the global optimization ability and local exploration ability,avoiding the premature phenomenon of the algorithm.Based on these improvements,the MCh OA and ICh OA are combined with the traditional Fuzzy Kapur objective function to perform threshold segmentation on benchmark images.After strict data and visual comparisons,it is found that the ICh OA algorithm has more advantages than the MCh OA algorithm in image threshold segmentation.The ICh OA algorithm is then applied to medical image segmentation for multiple threshold segmentation simulations,recording Peak Signal to Noise Ratio(PSNR)and Feature Similarity Index Measure(FSIM)values and comparing them with other popular algorithms.The experimental results show that the improved ICh OA algorithm has better segmentation accuracy and effectiveness.(3)To address the problem of capuchin search algorithm easily falling into local optima,a chaotic elite reverse learning strategy is used to initialize the position of the capuchin,which leads to a higher-quality initial population.In the algorithm iteration process,to avoid falling into local optima too early,we introduce Levy Flight disturbance strategy to form an Improved Capuchin Search Algorithm(ICap SA)with multiple strategies.Similarly,the improved ICap SA algorithm is combined with the Fuzzy Kapur objective function to perform multiple threshold segmentation simulations of medical images.PSNR and FSIM values are recorded,and the results are compared with those of other similar algorithms.The experimental results show that the improved algorithm can effectively enhance the visual effects of image segmentation.