The mammalian nervous system consists of a complicated network of biological structures, with functional subsystems constrained by a structural architecture that operates at scales spanning many orders of spatial magnitude. Our understanding of the architecture of the brain has been mediated through developments in biological imaging, though all imaging approaches are constrained by tradeoffs in achievable resolution, sensitivity, and field of view. Electron microscopy can be used to image nano-scale synapses, but only across small volumes, while magnetic resonance imaging (MRI) can be used to image whole brains but with spatial resolutions more coarse by several orders of magnitude. Developments in MR microstructural imaging methods such as diffusion MRI (dMRI) and echo-planar spectroscopic imaging (EPSI) help bridge the resolution and field of view gap by estimating cellular properties such as fiber orientations, myelin integrity, and long- range connectivity across the brain using clinically feasible acquisition sequences. These MR approaches rely on biophysical signal models to reconstruct sub-resolution properties of the underlying tissue. The theme of this dissertation is the development of tools and analysis methods used to perform multi-modal validation studies for these MR microstructural imaging models in the mouse brain, with specific focus on dMRI reconstructions and tractography. First, we demonstrate the utility of whole-brain synchrotron microcomputed tomography as a validation modality for the estimation of nerve fiber orientations with dMRI. MicroCT provides isotropic resolution across whole mouse brains with no physical sectioning, address- ing limitations in existing optical-based dMRI validation methods. Computer vision tools were developed to estimate fiber orientations that were spatially registered to dMRI data of the same specimen. Comparisons between modalities show good agreement in the rep- resentation of local fiber geometries and long-range trajectories, demonstrating the utility of synchrotron microCT for future dMRI validation studies. Furthermore, we show that microCT is compatible with follow-up electron microscopy, forming a multi-modal imaging pipeline capable of colocalizing structures across five orders of magnitude of resolution. Next, we perform statistical analysis with geometric surrogate graphs to explore the role of spatial embedding in the topological properties of the mouse structural brain network mea- sured with neural tracer imaging and dMRI tractography. We find that spatial embedding plays a considerably larger role in the topology of tractography networks than tracer net- works. Tractography underestimates long-range connectivity, which leads to geometric biases in the estimated modular structure and placement of hub nodes. Our results demonstrate the caution required in the interpretation of tractography-derived network measurements that rely on long-range connections and motivate additional geometric consideration in the design of future tractography validation studies. Finally, we analyze MR spectra from control and dysmyelinated mouse brain with EPSI to reveal limitations in existing biophysical compartmental models traditionally used for myelin imaging. We show that spectra estimated from these biophysical models fail to accurately predict the extent of asymmetric broadening in white-matter voxels, leading ultimately to compromised sensitivity to important differences in white-matter structure.Throughout, we highlight the value that high-resolution ground-truth imaging brings towards an understanding of the nature of the MR reconstruction problems themselves.