Diffusion MRI reveals amyloid plaque-associated cellular signatures through spatial transcriptomic integration.
Authors
Affiliations (7)
Affiliations (7)
- Advanced Imaging Research Center, UT Southwestern Medical Center, Dallas, USA.
- Department of Anatomy, Cell Biology & Physiology, Indiana University, Indianapolis, USA.
- Department of Neurological Surgery, Indiana University, Indianapolis, USA.
- Department of Surgery, University of Minnesota, Minneapolis, USA.
- Department of Biostatistics and Health Data Science, Indiana University School of Medicine, Indianapolis, USA.
- Department of Biomedical Engineering, UT Southwestern Medical Center, Dallas, USA.
- Peter O'Donnell Brain Institute, UT Southwestern Medical Center, Dallas, USA.
Abstract
Amyloid plaque deposition is a defining feature of Alzheimer's disease (AD), yet how plaque-associated cellular alterations are reflected in noninvasive imaging remains unclear. We integrated high-resolution diffusion magnetic resonance imaging (dMRI) with spatial transcriptomics (ST) in an AD mouse model. Spatial correlations were assessed across the whole brain and in disease-control and plaque-non-plaque comparisons. Neural network models were trained to predict AD-associated cell types from dMRI metrics. dMRI metrics showed moderate spatial correlations (r = -0.41 to 0.36) with oligodendrocyte- and neuron-related gene expression. Disease-control comparisons revealed region-specific MRI-derived microstructural alterations aligned with transcriptional changes in microglia, astrocytes, oligodendrocytes, and neurons. Plaque-rich regions exhibited distinct diffusion and susceptibility signatures associated with oligodendrocyte and microglial expression. Neural networks predicted spatial distributions of AD-associated cell types with 80%-86% accuracy. dMRI captures plaque-associated cellular pathology and supports biologically informed interpretation of imaging contrasts in AD.