Intrinsic Brain Activity Abnormalities In Spinal Cord Injury: A Resting State FMRI Study

Objective: To evaluate the change of functional connectivity andspontaneous neuronal activity of brain using independent resting-statecomponents(ICA)and amplitude of low-frequency fluctuation,(ALFF) ofresting state fMRI in spinal cord injury (SCI).Methods:22paraplegic patients,16tetraplegia patients and23age-matched healthy control subjects attended in this study. The imagingacquisition was done using a Siemens1.5-Tesla scanner while the subjectswere asked to keeping still. The resting-state fMRI BOLD images wereacquired axially using the following parameters:2000/27ms (TR/TE),20slices,4.0/1.0mm (thickness/gap),240mm×240mm (FOV),64×64(resolution),90o (flip angle). The resting-fMRI scanning session lasted for390s. After the data was collected, AFNI was used for data pre-processing whichinclude slice time correction; motion correction; spatial smoothing; filtering;removal of linear trends; spatial normalization. Preprocessed functional datafor all of the subjects were temporally concatenated, and then decomposedinto18independent components using Multivariate Exploratory LinearOptimized Decomposition into Independent Components (MELODIC). Theeight resting-state networks estimated using MELODIC were visuallyidentified by comparing with previously defined maps. Group-ICA spatialmaps were back-projected for each subject's fMRI dataset in order to estimatesubject-specific temporal components and associated spatial maps using thedual regression technique. For each group ICA component, AFNI was used toperform voxelwise one way Analysis of Variance (ANOVA) in order to assessany difference among the three groups of network strength maps obtainedfrom the three subject groups. Finally, for each ICA component, theprobability maps representing difference among the three groups were False Discovery Rate (FDR) corrected using AFNI. For each ICA component, theback projected spatial maps yielded by dual regression represented thesubject-level network strength of the brain network corresponding to each ofthe ICA component. For the two SCI groups–paraplegic and tetraplegic,Spearman correlation coefficients were calculated between theneuropsychological measurements and the mean network strength of theregions that exhibited group difference. This analyses was carried out usingMATLAB. ALFF changes in SCI patients also were investigated usingAFNI software.Result:(1) Eight major resting-state networks (Figure1)–visual, lateraloccipital, auditory, sensory motor, default mode, executive control and leftand right lateralized fronto-parietal-were visually identified by comparing the18ICA components yielded by MELODIC with previous reports.(2) Theright lateralized fronto-lateral network (RLFL) and the default mode network(DMN) showed significant differences among the three groups. Two regionsin the RLFL showed significant difference. Using the Talairch and Tournouxatlas the two regions were identified to be in the right inferior frontal gyrus(RIFG) near right precentral gyrus (X=-60, Y=-5, Z=25, Figure2A) andthe right cerebral white (RCW) matter near the right cingulate gyrus (X=-23,Y=13, Z=32, Figure2C), respectively. Unpaired t-tests between pairs of thethree groups indicated that the network strength within RIFG of tetraplegicgroup was significantly lower than that of the control and the paraplegicgroups (Figure2B). Unpaired t-tests between pairs of the three groups alsoindicated that the network strength in the RCW was significantly lower for thecontrol group as opposed to the two SCI groups (Figure2D). The rightanterior cingulate cortex (ACC) near Broadmann areas24and33(X=-4, Y=-20, Z=23, Figure2E) in the DMN was found to be significantly differentamong the three groups. Unpaired t-tests between pairs of the three groupsindicated that the network strength in ACC was significantly higher for thetetraplegic group compared to that of the control and the paraplegic groups(Figure2F).(3) the paraplegic group scored significantly higher than the tetraplegic group in the three behavioral tests.(4) as shown in Figure3, thecorrelation of the neuropsychological performances and the network strengthof the regions showing difference among the groups.(5) ALFF: PPG and TPGshowed higher ALFF than CTR in the insula, cingulate cortex (left and right)..Conclusion: This resting-state fMRI study suggests the resting statefunctional abnormalities of brain in SCI including the functional connectivityand spontaneous neuronal activity, which related to the level and seriousnessof the injury.