The Research On Key Techniques In MRM Based Small Animal Brain Image Processing

The brain is one of the most complex structures in the world,which is the central system that controls the human mind and activity.With the rapid development of the transgenosis and gene knockout techniques,more researchers focused on the study of brain science and translational medicine using animal models.Magnetic resonance microscopy(MRM)has the ability to capture the mouse brain 3D images in non-invasively way,which developed on the basis of magnetic resonance imaging(MRI)with high magnetic field and high resolution.The neuroimaging analysis of mice brain MRMs may give researchers a deep understanding of the mechanism brain diseases,brain aging and et al.However,some key technologies in mouse brain MRM analysis have not been fully covered,such as registration,segmentation and et al.In this study,we focus on three key techniques of the mouse brain MRM analysis,trying to solve these technical challenges and to make contributions to the development of the mouse brain MRM research.The contents of this research are as follows:(1)Research on optimization and evaluation the registration algorithms for mouse brain MRM.Our research based on five C57BL/6J mice T2 MRMs which was acquired using the same method.First,we have studied on the reliability of data individual brain atlas.Then,we specially tuned five diffeomorphic image registration algorithms(DARTEL,geodesic shooting,diffeo-demons,SyN(Greedy-SyN and geodesic-SyN))for mouse brain MRMs and evaluated their performance using three measures(volume overlap percentage(VOP),residual intensity error(RIE)and surface concordance ratio(SCR)).Our research certifications that geodesic-SyN was the best algorithm for mouse brain MRM registration.(2)Research on the segmentation of brain tissue structure in mice.This chapter mainly study around the mouse brain MRM simulation,in order to establishment a correspondence between anatomical atlas and mouse brain MRM.So far,the "gold standard" of the brain tissue segmentation is still the anatomical Brain Atlas,both in the brain and in mice.This study presents a three-dimensional texture synthesis algorithm for mouse brain MRM and a whole set of mouse brain MRM simulation process.In order to evaluation the simulation results,we have registered the simulation MRM to the raw MRM using two registration algorithms.Then,use the three evaluation criteria in the previous chapter for quantitative assessments.The research proves that the 3d texture synthesis algorithm we proposed can satisfy the needs of the mouse brain MRM simulation,and the simulation process is effective.(3)Study on the construction of individual structure network based on gray level distribution of mouse brain MRM image.The Brain network analysis is a new method of complex system analysis from the perspective of network.The construction of the brain network can be closer to the real topology of the brain,through the analysis of network topology properties can reveal this unusually complex system of internal structure of the work mode,and the network analysis will one of the most important methodes in neuroimaging research.This study developed on use Kullback-Leibler_divergence to construct invividual structure network.The results showed that the mouse brain MRM individual network based on Kullback-Leibler_divergence have small world properties.This proves that the method of constructing individual grey matter brain network based on Kullback-Leibler_divergence can be applied to the analysis of mouse brain.