White Matter Analysis Based On Spatial Priori Bayesian Model

Attention deficit hyperactivity disorder(ADHD),commonly known as ADHD,is a neurological disorder caused by abnormal white matter.It occurs mostly in childhood,and the clinical symptoms of some children will last until adulthood.In the field of brain imaging neurobiology,diffusion tensor imaging(DTI)can provide white matter information of the brain,and the use of DTI technology to explore white matter abnormalities in adult ADHD is gradually increasing.The main method of exploring white matter anomalies is to fit DTI data with diffusion tensor to obtain scalar information,and to measure the difference of white matter tissue in brain regions through scalar information.The quality of the diffusion tensor fitting method directly affects the integrity of the extracted scalar information,and then affects the difference statistics.The commonly used white matter analysis method is based on the traditional least squares diffusion tensor fitting method,which is susceptible to noise during the fitting process and has low accuracy.In order to overcome the shortcomings of commonly used analysis methods,a white matter analysis method based on spatial prior Bayesian model is proposed.Firstly,the DTI data containing white matter information is preprocessed by using the magnetic resonance image processing software package FSL.Secondly,the diffusion tensor fitting method based on spatial prior bayesian model is used to conduct tensor fitting on the preprocessed data to generate scalar information brain map.Next,using the TBSS module in the FSL to calculate the average skeleton of the white matter fiber bundle by scalar brain map for differential statistical analysis.Finally,the scalar information is extracted from the analysis results,and the scalar information is used as the feature for SVM classification.In tensor fitting,the spatial coherence of adjacent voxels is used to reduce noise,reduce uncertainty and improve accuracy.The DTI data of adult ADHD patients were selected as the research object,and compared with the traditional white matter analysis method,the validity of the proposed white matter analysis method was verified.The results show that the average skeleton information of white matter fiber bundles generated by the proposed white matter analysis method is more detailed,and the difference of scalar brain maps is more significant.At the same time,according to the statistical results of the differential brain regions obtained from the average skeleton information,under the traditional white matter analysis method,the FA value,MD value and RD value in the corpus callosum and the right front radial coronary region are selected as features for classification.The classification accuracy rate reached 67%.Under the white matter analysis method in this paper,the FA value,MD value,RD value and AD value in the corpus callosum,bilateral frontal radiation crown and right upper longitudinal bundle brain were classified as features,and the classification accuracy was 74%.Compared with the traditional method,it has increased by 7 percentage points.