Analysis Of MR Image And Research Of Related Issues For Computer-aided Diagnosis Of Infantile Cerebral Palsy

The incidence of infantile cerebral palsy disease is becoming higher and higher.At present,the research of computer aided diagnosis of infantile cerebral palsy disease is still on the stage of theoretical research. With rapid development of computer technology,computer aided diagnosis method has become an important research field in a medical image, diagnostic radiology, computer science.Research shows that using computer aided diagnosis method has played a positive role to improve the diagnostic accuracy and reduce misdiagnosis.Doctors can assessment and diagnosis of medical image according to the result of computer measurement.Using the computer to complete the division of pediatric brain tissue and lesion area extraction is the foundation of the auxiliary diagnosis of infantile cerebral palsy disease.This thesis analyzes the enhancement, segmentation method of the pediatric cerebral palsybrain image correlation, and corpus callosum disease region extraction and segmentation of the pediatric cerebral palsy.This thesis analyzs how to extract relevant brain areas in the following experiment.(1)Denoising methods of pediatric MR brain image are studied.Based on the partial differential equation, using a continuous threshold connects the two kinds of characteristic value correction algorithm based on structure tensor in succession, comprehensive utilization of both advantages of two algorithms to deal with the noise.(2) Pediatric brain tissue segmentation algorithm is studied.This article uses improved markov random field model based on EM algorithm for pediatric brain tissue segmentation,and introduces gaussian kernel nonparametric estimation in this model, etc. Experiments show that the algorithm is effective to reduce the dependence on brain atlas registration technology. (3)Corpus callosum lesions region extraction and segmentation associated with diseases of cerebral palsy are proposed in this thesis. Two kinds of interactive image segmentation algorithm are used, improved GraphCut and improved GrowCut. First of all, This thesis introduces the super pixels, nonparametric estimation to GraphCut algorithm.At the same time, according to the tedious process of the algorithm that twice selecting seed point is required on the fuzzy boundary, this thesis introduces the GrowCut algorithm, adds the fuzzy connective degree and the area constraint information to the algorithm.Experiments show that the algorithms obtain good segmentation effect.On the similarity coefficients(DICE), segmentation accuracy of the algorithm of this thesis is increased by 6%-7% than that of traditional GrowCut algorithm.Through experiments testing, the algorithm of infant brain image segmentation proposed in this thesis can fast and accurate segment brain tissue, while can reduce the dependence on brain atlas registration. The segmentation extraction algorithm of the corpus callosum structure based on interactive, compared with the traditional segmentation extraction algorithm,the algorithm of this thesis has greatly improved in segmentation efficiency and segmentation result,has practical application value.