To measure the brain-wide vascular density (VD) alteration in degenerated brains with Alzheimer's Disease (AD), deep learning-based super-resolution approach was developed to assist the segmentation of micro-vessels from the Monocrystalline Iron Oxide Nanoparticle (MION)-based CBV MRI images of transgenic mouse brains. Iron-induced T2* amplification effect well separated micro-vessels with tens of microns from capillary-enriched parenchyma voxels, enabling vascular compartment-specific VD differential analysis between AD and wildtype control mice. The differential maps based on segmented micro-vessels identified decreased VD in the anterior cingulate cortex (ACC) and medial entorhinal cortex (mEC) and increased VD in several highlighted brain regions, including dentate gyrus (DG) of the hippocampus, central and geniculate thalamus, medial septal area (MS), ventral tegmental area (VTA), and lateral entorhinal cortex (lEC). In contrast, the T2*-weighted capillary density mapping from parenchyma voxels showed increased VD in several cortical regions, including somatosensory and visual cortex, retrosplenial cortex, as well as piriform area and lEC in AD brains. However, dramatic capillary VD decrease was observed in the subcortical areas including hippocampus, thalamus, hypothalamus, and pontine areas. These high-resolution MION-based CBV MRI elucidates altered vascular compartments in degenerated AD brains, reconciling the various region-specific vascular impairment and angiogenesis in functional areas critical for cognitive decline of AD.