| Objective:To explore the optimized diffusion approach in detecting the microstructural changes using the non-Gaussian diffusion model to fit the spatial characteristics of the tissue. And to evaluate voxel-based white matter (WM) and gray matter (GM) non-Gaussian diffusion changes of the whole brain in early Alzheimer’s disease (AD) patients, using diffusion kurtosis imaging (DKI).Subjects and Methods:6 health right-handed volunteerswere recruited to have the conventional MRI and diffusion sequence. The basic reference b=0 as well as b=500s/mm2, b=1000s/mm2, b=1500s/mm2, b=2000s/mm2, b=2500s/mm2 were used in diffusion sequence. Diffusion gradients-encoding directions were 30; Preprocession was conducted using the "eddy-current" toolbox in FSL. Maps were calculated in the MATLABR2010b platform, including apparent kurtosis coefficient (AKC),apparent diffusion coefficient (ADC), axial kurtosis (AK) and radial kurtosis (RK). Then the contrast among GM, WM and cerebrospinal fluid (CSF), as well as the microstructual imaging of the corpus callosum (CC) were evaluated.Then 26 early AD patientsand 26 normal controlswere recruited. Basic information and the related clinical history were collected. These subjectsalso underwent the neuropsychological test, including the Mini-mental state examination (MMSE), Montreal cognitive assessment (MOCA), clinical dementia rate (CDR), Hachinski Ischemic Score (HIS), Hamilton Depression Scale (HAMD) and Hamilton Anxiety Scale (HAMA). Convention and diffusion data were acquired andunderwent the preprocession. Seven parametric maps were calculated in the DKE software. All theseDKI indexes, including mean kurtosis (MK), AK, RK andfactional anisotropy(FA), mean diffusivity(MD), axial diffusivity(AxD),radial diffusivity(RD)of the whole brain underwent the voxel-based analyses (VBA). Correlationanalysis between the DKI and the cognitive performance scores were performed.Results:1① Increasing the number of b-value only cannot improve the kurtosis contrast of different tissues, and may even increase the model fitting error as well as the acquisition time. And the combination of 2 b-value is more suitable for clinical application;② The combination of 1000s/mm2 and 2000s/mm2 can detect the micro structural characteristics of the corpus callosum more truthfully.2①In WM of the AD group, CC, bilateral cingulate bundle, bilateral frontal WM and temporal WMsignificantly increased in MD, AxD, RD and decreased in FA and MK, AK, RK(P<0.001);②In GM of the AD group, parahippocampal gyrus, hippocampus, and cingulate gyrus, subcallosal gyrus,bilateral frontal GM and temporal GM showed significantlyincreased MD, AxD and RD, as well as significantlydecreased MK, AK and RK (P<0.001).However, bilateral lentiform nucleus in the subcortical area increased in kurtosis indexes;③Abnormalities in AD that found by MK, AK and RK were differentfrom that byFA and MD, AxD, RD (P<0.001);④MK in CC, left cingulate bindle, bilateral frontal WM and bilateral temporal WM correlated positively with the clinical performance scores (P<0.001).FA in CC and left cingulate bindle correlated positively with the clinical performancescores. MD in CC and left cingulate bindle correlated negatively with the clinical performance scores. MK detected more correlated regions than FA and MD.Conclusion:The combination b-value of 1000s/mm2 and 2000s/mm2 is appropriate to get the optimal DKI sequence of the central nervous system. Early AD patients already have abnormalities on both WM and GM, and have increased complexity in lentiform nucleus. All these abnormalities show non-Gaussian diffusion alteration, and can be detected by DKI. Additionally, the indexes in different regions were correlated with the clinical performance. This indicated that different indexes of DKI had varied sensitivity in detecting the different diffusionalteration. DKI can provide better biomarkers in reflecting AD severity and may be a powerful compensation in diagnosing early AD than conventional diffusion tensor imaging (DTI). |
PDF Full Text Request