MAKING BRAINGLOBE MORE ACCESSIBLE: SUPPORTING PARTIAL BRAIN REGISTRATION AND ACCELERATING HIGH-THROUGHPUT CELL DETECTION

The BrainGlobe Initiative was created to facilitate development of Python-based tools for computational neuroanatomy, providing interoperable software accessible across different platforms and compatible with multiple model organisms. At present, we maintain 18 packages with 100+ contributors, fostering a community of neuroscientists and developers to share knowledge, build software, and engage with the scientific and open-source community.

One key limitation of the BrainGlobe ecosystem has been the lack of flexibility in the type of data that can be registered using currently available BrainGlobe tools. The ecosystem relies on brainreg to transform data into a common coordinate space, enabling standardised anatomical comparisons across studies. However, brainreg required 3D whole-brain data as input, excluding researchers working with partial brain samples. To address this limitation we developed brainglobe-registration, which can register 2D slices, 3D sub-volumes, and whole-brain data into a BrainGlobe atlas.

In parallel with the registration improvements, we enhanced cellfinder, a tool that automates cell detection in large 3D volumes. With updates to the data loading and pre-processing steps, cellfinder is now 3-5x faster at processing data from experimental paradigms such as whole brain c-Fos labelling, which can generate millions of labelled cells.

Together, the recent work expands accessibility to the BrainGlobe ecosystem for researchers working with partial brain samples and dramatically accelerates the analysis of large-scale whole-brain cell labelling datasets.