Research On Algorithm For Medical Image Segmentation Based On Fuzzy Cluster And Cuckoo Optimization

With the rapid development of medical imaging technology,the use of computer-aided diagnosis for medical image segmentation has gradually become more and more popular.However,there are many problems in the segmentation of medical images at present,such as Magnetic Resonance Imaging(MRI)noise interference,uneven radio frequency intensity,and local volume effect,etc.,which will affect the accuracy of segmentation and even affect the doctor’s judgment of the disease.In this paper,by improving the traditional cuckoo algorithm,using its excellent global optimization characteristics to optimize the cluster center of the fuzzy clustering algorithm,and introducing an anti-noise factor in the objective function,it improves the algorithm’s global search ability in medical image segmentation.While improving the convergence speed of the algorithm,it also suppresses noise.The main research work is as follows:1)Aiming at the local search ability of traditional cuckoo algorithm is not good enough and the parameter setting is fixed,a cuckoo search algorithm based on Sine map and simplex method is proposed,and the discovery probability Pa is improved,using sinusoidal adaptive discovery Probability to optimize the accuracy of the algorithm.In the simulation experiment,the improved algorithm and other six-species intelligent algorithms are used to conduct optimization experiments under 10 test functions(including unimodal and multimodal).Experimental results show that the improved algorithm is superior to other algorithms in terms of optimization accuracy and convergence speed.2)Aiming at the shortcomings of fuzzy clustering algorithm in segmenting brain MRI images that the global search ability is weak and easy to fall into the local optimum,a fuzzy clustering algorithm based on improved cuckoo(ICS＿FCM)is proposed,which uses the powerful improved cuckoo algorithm The global optimization feature optimizes the clustering centers,recodes and designs bird nests and cuckoo populations,and segments the T1-weighted cross-sectional,sagittal,and coronal images of brain MRI.Experimental results show that the segmentation effect of the improved algorithm is better.3)Aiming at the problem of fuzzy clustering algorithm’s sensitivity to medical image noise and slow convergence speed,this paper introduces neighborhood spatial information into the algorithm’s objective function,and proposes an improved fuzzy clustering image segmentation algorithm with anti-noise(ICS＿FCM＿S1),Use the improved algorithm to segment brain MRI T2-weighted brain hemorrhage,brain calcification,meningioma and other six kinds of brain disease images,and combine with the active contour model to extract the region of interest in the image.The results show that the improved algorithm has achieved a good segmentation effect,and the noise resistance has also been improved.In short,this article improves the traditional cuckoo algorithm and fuzzy clustering algorithm,improves the algorithm’s optimization and convergence performance,and optimizes the noise sensitivity and slow convergence in medical image segmentation,and improves the accuracy of the algorithm segmentation.,To provide references for future research in the fields of computer-aided diagnosis and disease screening.