Study On Hydrocephalus Computer Aided Diagnosis Based On Ventricle Segmentation

Nowadays,more and more brain diseases have affected people’s life,hydrocephalus,as one of them,has brought great harm to people’s health.So timely diagnosis of hydrocephalus is very important.In traditional method,doctors do diagnosis by reading the MR images.Along with the rapid increment of medical image data,this method is time consuming and has low effectiveness.Meanwhile,the rapid development of medical imaging technology,such as magnetic resonance,has increased the demand for computer-aided diagnosis in clinical diagnosis greatly,in which,medical image segmentation is an important part of diagnosis.Thus,the brain medical image segmentation becomes one of the hotspots in the cross-subject field of Computer and medical science.The obvious expansion of the lateral ventricle is an important feature of hydrocephalus,the accuracy of lateral ventricle segmentation is directly related to the effect of subsequent hydrocephalus computer-aided diagnosis,so the main content of this thesis is to research the lateral ventricle segmentation technology of medical image processing.Firstly,the significance of medical image segmentation and lateral ventricle segmentation and research status at home and abroad will be introduced.Then,the brain MR image and its feature and the structure of the lateral ventricle will be introduced.At last,some related medical image segmentation and medical image processing algorithms will be studied in this thesis,on this basis,some of them will be improved,and the lateral ventricle segmentation model will be proposed.Specifically,the main research works are as followings:1)The related brain medical image segmentation technology will be summarized,the principle and content of different algorithms will be introduced and compared in this thesis.2)The related segmentation algorithms that can be used in lateral ventricle segmentation will be studied and the advantage and disadvantage of them will be analyzed.3)The region growing algorithm and threshold segmentation algorithms will be deeply researched in this thesis.According to the research content of the thesis,the 3D region growing algorithm and OTSU algorithm will be improved.Meanwhile,morphological operations and other image processing methods will be adopted during the process.On this basis,lateral ventricle segmentation combination model will be proposed.4)Based on the multiple maximum between-cluster variance proposed in the lateral ventricle segmentation combination model,the cranial cavity segmentation scheme will be constructed in this thesis.The hydrocephalus computer-aided diagnosis system will be realized based on the segmentation consequence of lateral ventricle and cranial cavity.In order to show the efficiency and accuracy of the lateral ventricle segmentation combination model proposed in the thesis,contrast experiments between the proposed model and some other representative methods on the simulated and clinical medical images will be done.