Research On MRI Image Segmentation Based On Fuzzy C-Means Clustering Algorithm

Brain MRI image segmentation is based on the similarity between MRI and a certain standard.Pixels with high similarity in brain MRI are divided into the same tissue class,and conversely,they are divided into different tissue classes.In clinical medicine,the MRI image of the brain makes the image itself uncertain and fuzzy due to its unique imaging method,and the high complexity of the brain tissue structure makes it difficult to segment the image.Improving the efficiency of clinical diagnosis and adjuvant treatment is the top priority of research.FCM algorithm based on fuzzy theory is currently one of the most suitable methods for brain MRI segmentation.Many researchers have studied and improved the FCM algorithm.Based on this,the following research is done in this article:(1)The traditional FCM algorithm and the improved algorithm using the spatial information of the image are studied.(2)This paper proposes an improved fuzzy C-means clustering algorithm based on non-local spatial information.The algorithm first uses non-local means to denoise the image using the non-local spatial information of the original image;secondly,uses the histogram to detect the optimal number of segmentation categories;finally,it introduces a blur factor into the objective function,and this blur factor is fully utilized Local spatial information of the original image.The algorithm finds a balance between the image detail information and the robustness of noise.The segmentation effect diagram clearly shows that the algorithm has a good segmentation result for noisy images.(3)Use MATLAB tool to realize the segmentation test of noisy brain MRI proposed by FSICM algorithm,and compare with FCM,KFCM,FCM-S1,FLICM and FCM-NLS algorithms.According to the F-score evaluation index rate,the algorithm and experimental results are evaluated and analyzed.The results show that when the algorithm proposed in this paper segmentes high-noise brain MRI images,the background segmentation accuracy is99.76% and the cerebrospinal fluid segmentation accuracy is 86.5%.The accuracy of brain gray matter segmentation is 92.26%,and the accuracy of brain white matter segmentation is97.76%,which is extremely accurate and resistant to noise.