ObjectivesTo investigate brain microstructural changes in white matter and gray matter of type 2 diabetes mellitus (T2DM) patients using diffusion kurtosis imaging.MethodsDiffusion kurtosis imaging (b values=0, 1250, and 2500s/mm(2)) was performed for 30 T2DM patients and 28 controls. FMRIB Software Library with tract-based spatial statistics was used to analyze intergroup differences in fractional anisotropy (FA), mean diffusivity (MD), mean kurtosis (MK), axial kurtosis (K-vertical bar), and radial kurtosis (K) of multiple white matter regions. Atlas-based ROI analysis was conducted in gray matter structures and some fiber tracts. Correlations between MK changes and clinical measurements were determined.ResultsIn whole-brain tract-based spatial statistics analysis, T2DM patients exhibited abnormalities in 29.6%, 30.4%, 35.4%, 10.5%, and 26.0% of white matter regions as measured by FA, MD, MK, K-vertical bar, and K, respectively, when compared to the controls. MK reduction was contributed more by the decreased K. In atlas-based analysis, MK detected more ROIs (27/48) with white matter microstructural changes than FA (13/48) and MD (17/48). MK decreased in bilateral thalamus and caudate, while FA showed statistically significant difference only in the left caudate. MK values negatively correlated with disease duration in the genu of corpus callosum and anterior corona radiata (R=-0.512 and -0.459) and positively correlated with neuropsychological scores in the cingulum (hippocampus) (R=0.466 and 0.440).ConclusionsDiffusion kurtosis imaging detects more brain regions with white matter and gray matter microstructural alterations of T2DM patients than DTI metrics. It provides valuable information for studying the pathology of diabetic encephalopathy and may lead to better imaging biomarkers for monitoring disease progression.