Initial Correlation Analysis of Diffusion Basis Spectrum Imaging of Alzheimer's Brain and Quantitative Histology

Abstract

AbstractBackgroundDiffusion Basis Spectrum Imaging (DBSI), a novel MR imaging method, can be utilized as a biomarker in Alzheimer's disease (AD) staging and progression. By distinguishing diffusion effects from other concomitant disorders like axonal injury/loss and demyelination, it is capable of specifically evaluating cellularity and edema in AD brains. To test the hypothesis that DBSI cell fraction and mean cell apparent diffusion coefficient (ADC) correlate with measures of cell density and neuroinflammation, we employed high‐resolution ex vivo DBSI sequences and quantitative digital histology on post‐mortem human AD brain tissue.MethodA 15 × 20 × 6 mm3 sample of middle frontal gyrus (coronal plane) from a formalin‐fixed post‐mortem human AD brain (Knight ADRC) was incubated in phosphate‐buffered saline for 48 hours and then imaged with a Varian 11.7T MRI, using 2D spin‐echo sequence with 20‐direction diffusion encoding (max b = 3900 s/mm2, voxel = 0.25 × 0.25 × 1 mm). After MRI, it was precisely bisected along the plane of imaging, and processed into paraffin. Slide‐mounted 6 µm histologic sections were stained with H&E and DAB immunohistochemistry (IBA‐1 antibody, hematoxylin counterstain) and digitized (20× magnification, Hamamatsu NanoZoomer). Maps of nuclei and microglia density were computed by segmenting nuclei and microglia respectively using Orbit Image Analysis and in‐house MATLAB programs. These histology maps were registered and downsampled to the corresponding DBSI maps. We generated cell fraction maps and mean cell ADC maps from MR images using in‐house DBSI software by computing the fraction and mean ADC of restricted isotropic diffusion (0 – 0.4 ×10‐3 mm2/s in each image voxel, respectively.ResultAmong 71 evenly spaced regions of interest (ROIs, 2 × 2 mm2) drawn across the entire specimen, the mean DBSI cell fraction correlated with nuclei density (R = 0.74, P < 0.001). Additionally, among 42 ROIs drawn in the white matter region, the mean cell ADC in each ROI correlated with microglia density (R = 0.54, P < 0.001).ConclusionOur findings demonstrated the feasibility of DBSI technique in ex vivo AD brain, which could improve our understanding of neuroinflammation in AD pathogenesis, early diagnostics and new disease‐modifying therapies targeting the immune response.