Application Of Brain Anatomical Connection Analysis Based On Fiber Bundle Clustering

At present,the traditional voxel-based analysis is widely used in the study of the anatomical properties of white matter fibers.Regions of interests are selected to compare and analyze the differences of brain area indicators between different samples.There are two limitations in this method.One is the defects of accuracy caused by sample registration,the other is that the average value of the index calculated by traditional analysis will lead to homogenization of fiber characteristics"and hide the global diffusion information.In order to overcome the shortcomings of traditional research,the white matter properties of complete fiber bundles are analyzed,and the time scale problem of increasing fiber tracking data is optimized.Fiber based analysis was selected to classify white matter automatically by two clustering methods:regional clustering and atlas clustering.The diffusion characteristics of different groups along the fiber bundle and the variation of anatomical properties were calculated.The specific work and achievements of this paper are as follows:Firstly,aiming at the defect of traditional method that voxel index mean is used instead of experimental data,this paper identifies several main fibers of human brain by means of automated fiber quantization,and samples them equidistantly along the fibers,and calculates the indexes to mark local differences.In the experimental results,there were continuous significant differences on some fiber bundles;the diffusion characteristics of the whole fiber showed a curve-like increase or decrease;professional chess players also showed a certain increase in the diffusion index compared with the ordinary samples,and the Pearson related experiments with clinical indicators also confirmed this difference.Subsequently,in order to remedy the defect that clustering method only considers the central part of white matter,the brain network connection between the initiation and termination areas of individual significant fibers was detected,and the results were different,which strengthened the reliability of the conclusion and confirmed the changes of white matter anatomical connection in some areas of professional chess players.Secondly,considering that the automated fiber quantization method is rough in segmentation,the data dimension is single,and the experimental results are inconsistent.In this paper,the data-driven clustering method of brain atlas is used to refine the brain area fibers,further study the structure of human brain fibers and anatomical indicators,and verify the feasibility of clustering algorithm to detect brain connectivity attributes.In the experiment,80 subclasses of the upper longitudinal tract with strong significance in the previous experiment were extracted from 800 classes of the whole brain cluster,and the anatomical indexes were compared and analyzed one by one.In the experimental results,significant points were detected in each part of the upper longitudinal bundle and showed continuity.This corresponds to the experimental results in Chapter 3 to a certain extent.At the same time,there is a clear distinction between the two groups of data on the density of fibers,which proves the growth of fibers under the change of brain function.In this paper,the white matter is segmented by fiber clustering,and the anatomical connection of the brain is analyzed.On the one hand,by calculating the significance between professional chess players and ordinary people,we can prove the influence of brain function changes on the level of anatomy.On the other hand,aiming at the shortcomings of traditional analysis methods,two different fiber clustering algorithms are proposed to identify the long fibers that control important functions in human brain,and quantify the diffusion information along the fibers point by point.Summarizing the experimental results of the two groups,the overall results confirm the differences in the performance of professional chess players in different areas along the bundle of fibers compared with the general population.In addition,long-term cognitive activities,such as chess,may affect white matter fiber properties related to early memory,attention and visual pathways.The validity of clustering method in labeling local white matter differences was confirmed,which provided a new idea for the study of brain white matter development.