摘要：Magnetic resonance imaging (MRI) has become animportant auxiliary means in the clinical diagnosis of braindiseases and other diseases. The brain images segmentationbased on MRI technique mainly includes the segmentation ofnormal brain tissues and brain images with lesions. Thegray-scale distribution of MRI brain images is not uniformbecause of the noise, migration field effect and partial volumeeffect in the process of producing the MRI images, and theperformance under different image segmentation algorithms isdifferent. The segmentation experiments on single tissue images,MRI brain images and brain abnormal tissue images werecarried out based on seven segmentation algorithms (level set,cross entropy, fuzzy entropy, maximum between-class variance,expectation maximization, k-means clustering and fuzzyc-means clustering). The simulation results compared theperformance of these seven image segmentation algorithms onthe typical MRI brain images.
