Imaging Biomarker for Early-Stage Alzheimer Disease: Utility of Hippocampal Histogram Analysis of Diffusion Metrics

BACKGROUND AND PURPOSE: Biomarkers have been required for diagnosing early Alzheimer disease. We assessed the utility of hippocampal diffusion parameters for diagnosing Alzheimer disease pathology in mild cognitive impairment. MATERIALS AND METHODS: Sixty-nine patients with mild cognitive impairment underwent both CSF measurement and multi-shell diffusion imaging at 3T. Based on the CSF biomarker level, patients were classified according to the presence (Alzheimer disease group, n = 35) or absence (non-Alzheimer disease group, n = 34) of Alzheimer disease pathology. Neurite orientation dispersion and density imaging and diffusion tensor imaging parametric maps were generated. Two observers independently created the hippocampal region of interest for calculating histogram features. Interobserver correlations were calculated. The statistical significance of intergroup differences was tested by using the Mann-Whitney U test. Logistic regression analyses, using both the clinical scale and the image data, were used to predict intergroup differences, after which group discriminations were performed. RESULTS: Most intraclass correlation coefficient values were between 0.59 and 0.91. In the regions of interest of both observers, there were statistically significant intergroup differences for the left-side neurite orientation dispersion and density imaging–derived intracellular volume fraction, right-side diffusion tensor imaging-derived mean diffusivity, left-side diffusion tensor imaging–derived mean diffusivity, axial diffusivity, and radial diffusivity (P CONCLUSIONS: Hippocampal diffusion parameters might be useful for the early diagnosis of Alzheimer disease.