The Study Of Intracranial Hematoma Image Segmentation Method Based On Extension Detecting And FCM Clustering Algorithm

Human brain is the organ of human spirit and intelligence, the most advanced part of central nervous system, the center of all human activities, brain health is related to the quality of human life. Due to fights, traffic accident, and contretemps, the proportion of intraeranial hematoma to brain diseases is very high. Volume of intracranial hematoma, is an important basis for doctor to make clinical diagnosis and understanding disease state, while CT imaging examination is the main method for hematoma diagnosis, therefore implement of accurate automatic segmentation for CT images of intracranial hematoma region will lay the foundation of intracranial hematoma three-dimensional reconstruction and volume accurate computation of intracranial hematoma.This paper depthly investigates the development of medical image segmentation, present the new method of combining extension detecting matter focusing method and fuzzy C-means clustering algorithm,so as to realize automatic segmentation of intracranial hematoma region based on CT medical image. Matlab simulation experiment proves that this method can realize automaticly and accuratly segmentation of intracranial hematoma region based on CT medical image, and reach the neurosurgeon’s requirements.This paper studies extension detecting theory and applies extension detecting technology into the field of image segmentation. According to the problem of clustering center easy to fall into local optimum in FCM, this paper adopt the matter-focusing method in extension detecting theory to solve this problem, gives the related solution, and proves the effectiveness of the method by Mallab simulation.According to the characteristics of CT medical image of intracranial hematoma, this paper uses two-step segmentation methods to achieve hematoma region segmentation. The purpose of the first step segmentation is to remove the skull, by adopting threshold segmentation method and region growing method. The purpose of the second step segmentation is to segment hematoma region by adopting the new method proposed in this paper. When using fuzzy C-means clustering algorithm in image segmentation, if only using one single-pixel gray value as the characteristic value, misclassification phenomenons of noise regions will appear, In order to solve this problem, this paper adopts mean value of eight-neighboring pixels as the second characteristic value. Matlab simulation experiments show that this method can restrain noise very well and reduce the misclassification phenomenon effectivly.This paper has the following main innovations:Firstly, it’s the first time to propose to apply extension-detecting theory into the field of medical image segmentation, and establish the relevant matter-element model for intracranial hematoma; Secondly, it’s the first time to propose to use matter-focusing method in extension detection theory to solve the problem that cluster centers are easy to fall into local optimal in fuzzy C-means clustering algorithm; Thirdly, this paper successfully realizes intracranial hematoma region segmentation based on CT medical image by adopting the new method, and illustrates that the segmentation effect of this new method is superior to that of traditional fuzzy C-means clustering algorithm by the results of abundant simulation experiments.