High-resolution mri guided whole mouse brain neuronal cell type atlas using deep learning

Abstract

Cell types represent groups of cells with shared anatomical and functional properties. Traditional brain cell type atlases rely on single-cell sequencing, which provides molecular detail but lacks whole-brain, isotropic resolution. Diffusion magnetic resonance imaging (dMRI) offers a complementary approach for probing cytoarchitecture and myeloarchitecture, with quantitative metrics increasingly used as biomarkers of brain development and neurodegenerative disorders. However, the capacity of dMRI to predict cell types remains unclear. Here, we develop a multimodal framework by integrating high-resolution dMRI and three-dimensional light-sheet microscopy of adult mouse brains through registration to the Allen Mouse Brain Common Coordinate Framework. We investigate correlations between dMRI and spatial transcriptomics-derived cell types and generate a whole-brain neuronal cell type atlas at 10 µm isotropic resolution using deep learning. Together, these results establish an efficient, high-resolution strategy for brain neuron atlas generation and underscore the potential of advanced imaging techniques to illuminate cellular mechanisms of the brain.