Voxel-based Analysis And Multivariate Pattern Analysis Of Diffusion Tensor Imaging In Anti-NMDA Receptor Encephalitis

Objective The clinical manifestations of anti-N-methyl-D-aspartate receptor(anti-NMDAR)encephalitis are diverse,and cognitive dysfunction is one of the main clinical symptoms.But there are few researches on structural imaging and cognitive function of anti-NMDAR encephalitis.This study aimed to explore the brain white matter changes in patients with anti-NMDAR encephalitis by voxel-based analysis of diffusion tensor imaging,and whether the changes are related to cognition.Based on diffusion tensor imaging,multivariate pattern analysis(MVPA)was used to explore the feasibility of automatic classification of anti-NMDAR encephalitis and brain regions that play an important role in classification.Methods 15 patients with anti-NMDAR encephalitis and 15 healthy controls underwent Montreal cognitive assessment(MOCA)testing and diffusion tensor imaging scans.The white matter fractional anisotropy(FA)and average dispersion rate(MD)of the two groups were analyzed by voxel-based analysis,and the different white matter areas were extracted to study the relationship between white matter changes and cognition.Based on FA and MD for MVPA classification,the weights of each brain region were calculated.Results Compared with the controls,the patients showed an FA reduction in right middle temporal gyrus,left middle cerebellar peduncle,right precuneus,and an increase in left medial temporal gyrus and left frontal lobe.The MOCA score for patients was lower than controls,especially in executive function,fluency,delayed recall and visual perception items.The FA value of right precuneus was positively correlated with total MOCA score and fluency score.The MD of left frontal lobe was negatively correlated with total MOCA score,and MD of the left medial temporal gyrus was positively correlated with delayed recall.The accuracy,sensitivity and specificity of classification based on FA were 70%,60% and 80%,respectively.Based on MD,they were each 80%.The brain regions with large weights from FA and MD overlap in temporal lobe,cerebellum,and hippocampus.Conclusions These results suggest that multiple white matter brain structural damage and multiple cognitive domain impairments in patients with anti NMDAR encephalitis,and the changes in white matter are associated with cognitive deficits.MVPA based on FA and MD has a good classification ability.Diffusion tensor imaging and multivariate pattern analysis are helpful for us to better understand the pathophysiological mechanism of anti-NMDAR encephalitis.