Hierarchical Joint Registration of Tissue Blocks With Soft Shape Constraints For Large-Scale Histology of The Human Brain

Abstract

Large-scale 3D histology reconstruction of the human brain with MRI as volumetric reference generally requires reassembling the tissue blocks into the MRI space, prior to any further reconstruction of the histology of the individual blocks. This is a challenging registration problem, particularly in the frequent case that blockface photographs of paraffin embedded tissue are used as intermediate modality, as their contrast between white and gray matter is rather low. Here we propose a registration framework to address this problem, relying on two key components. First, blocks are simultaneously aligned to the MRI while exploiting the spatial constraints that they impose on each other, by means of a customized soft shape constraint (similarly to a jigsaw puzzle). And second, we adopt a hierarchical optimization strategy that capitalizes on our prior knowledge on the slicing and blocking procedure. Our framework is validated quantitatively on synthetic data, and qualitatively on the histology of a whole human hemisphere.