BIFROST: a method for registering diverse imaging datasets

Quantitative comparison of brain-wide neural dynamics across different experimental conditions often requires precise alignment to a common set of anatomical coordinates. While such approaches are routinely applied in functional magnetic resonance imaging (fMRI), registering in vivo fluorescence imaging data to ex vivo-derived reference atlases is challenging, given the many differences in imaging modality, microscope specification, and sample preparation. Moreover, in many systems, animal to animal variation in brain structure limits registration precision. Using the highly stereotyped architecture of the fruit fly brain as a model, we overcome these challenges by building a reference atlas based directly on in vivo multiphoton-imaged brains, called the Functional Drosophila Atlas (FDA). We then develop a novel two-step pipeline, BrIdge For Registering Over Statistical Templates (BIFROST), for transforming neural imaging data into this common space, and for importing ex vivo resources, such as connectomes. Using genetically labeled cell types to provide ground truth, we demonstrate that this method allows voxel registration with micron precision. Thus, this method provides a generalizable pipeline for registering neural activity datasets to one another, allowing quantitative comparisons across experiments, microscopes, genotypes, and anatomical atlases, including connectomes.