“…This paper introduces a new class of image-based diffeomorphometry methods which we term Projective LDDMM for aligning sparse sets of image captures to 3D coordinate systems across micron and millimeter scales. We focus particularly on the registration of 3D MRI with 2D digital histology, as representative of the class of multi-scale, multi-modality mapping in biomedical research including traditional light microscopy mapping to dense reference atlases [5, 6, 7, 8], light sheet methods [9, 10], deep tissue imaging [11], and spatial transcriptomics [12, 13, 14, 15]. We formulate the mapping of dense atlases to sparse images problem using the random orbit model of computational anatomy [16, 17, 18, 19] in which the space of dense anatomies is modeled as an orbit of a 3D template under the group of diffeomorphisms.…”