Study Of Brain Structural Network Properties And Texture Characteristics In Epilepsy Patients With Conventional MRI-negative

PART 1 Study of brain network properties of conventional MRI-negative epilepsy based on DTI sequenceObjective: Using magnetic resonance imaging of conventional MRI epilepsy patients with brain network at ribute analysis,exploring the structure of conventional MRI epilepsy patients with network connection change,explore conventional MRI negative or sensitive imaging characteristic of people with epilepsy tag,as the study of conventional MRI epilepsy patients with nerve pathological physiological mechanism provides new idea.Methods: Nineteen patients with conventional MRI negative epilepsy and 24 healthy subjects meeting the inclusion criteria were selected.All subjects underwent cranial axial 3D-T1 WI and DTI scanning using a GE 3.0T magnetic resonance scanner.General information of the patient was col ected,including age and sex.The magnetic resonance imaging data were preprocessed firstly,and the brain network matrix was constructed by using DTI sequence in PANDA software,and the attributes of the brain structure network were analyzed.Results: The brain structure network properties of the patients of MRI-negative and the healthy control group showed that both groups had small-world properties,and there was no significant difference in the small-world network characteristic parameters between the two groups.In terms of global attributes(P>0.05),the global efficiency of patients with negative conventional MRI decreased compared with the healthy control group,and the difference between the two groups was statistical y significant(P<0.05).There was no statistical y significant difference between the two groups in local efficiency,Assortativity,synchronization,and hierarchy(P>0.05).Conclusion: The brain network properties of conventional MRI-negative epilepsy have a small-world property compared with normal people,but the global efficiency is smaller than normal people in the global property.This may suggest that the global efficiency in the brain structure network of conventional MRI-negative epilepsy can be used as a neuroimaging marker,but further research is needed.PART 2 Study on image texture analysis of conventional MRI-negative epilepsy based on MRI multiple sequencesObjective: The texture characteristics of MRI images of patients with conventional MRI negative epilepsy were studied by using texture analysis method,and the characteristic texture parameters of conventional MRI-negative epilepsy patients were explored,so as to provide reliable information and objective basis for the diagnosis of conventional MRI-negative epilepsy patients.Methods: Nineteen patients with conventional MRI negative epilepsy and 24 healthy subjects meeting the inclusion criteria were selected.All subjects underwent transverse cranial T1-FLAIR,T2 WI,T2-FLAIR and DWI sequences using GE 3.0 T magnetic resonance scanner.Firstly,ITK-SNAP software was used to segment the T1-FLAIR,T2 WI,T2-FLAIR and DWI sequence images of conventional MRI negative epilepsy and normal heads artificial y.AK software was used to extract texture features,m RMR and Lasso regression methods were used to reduce the dimension of texture features,and multi-factor Logistic regression was used to construct the above four sequence texture feature models.Receiver operating characteristic curve(ROC)was used to compare the efficacy of the four models.Results: From T1-FLAIR,T2 WI,T2-FLAIR and DWI images can extract 1316 texture parameters,obtained by feature selection and dimension reduction 10,11,10,12 characteristic parameters,establish a set of texture feature model respectively,by the participants work curve(ROC)analysis,obtained based on T1-FLAIR,T2 WI,T2-FLAIR and DWI image feature extraction can establish the model of the AUC are 0.897,0.882,0.836,0.941,respectively.Hosmer-Lemeshow test showed that the models had good goodness-of-fit(all P> 0.05).Conclusion: The texture analysis based on Multi-sequence magnetic resonance imaging has a certain value in the identification of conventional MRI negative epilepsy from normal people,in which DWI sequence is the most effective in the identification of texture feature models,suggesting that texture feature can be used as neuroimaging markers for conventional MRI negative epilepsy,but further sample research is needed.