XClose

Advanced Research Computing

Menu

BrainGlobe Template Builder

Atlas building is complex and time-consuming. To help the BrainGlobe team meet growing demand, we streamlined the first stage: generating a high-quality, standardised 3D reference template.

Coronal view of blackcap brain template with atlas annotations overlay on the right.

11 May 2026

ARC has contributed to a number of BrainGlobe projects over the years. The BrainGlobe Initiative focuses on providing the research community with open‑source, interoperable tools for computational neuroanatomy. The initiative is led by the Neuroinformatics Unit (NIU) at the Sainsbury Welcome Centre (SWC) and continues to grow through contributions from researchers and developers across the community. 

For our latest BrainGlobe project, we worked closely with Alessandro Felder and the rest of the BrainGlobe development team to enhance BrainGlobe Template Builder.  

The first digital, high‑resolution 3D bird brain Atlas 

Most atlases available in BrainGlobe were originally created by external research labs using their own imaging platforms andatlas building workflows. After publication, these atlases are converted into the BrainGlobe format so they can be used within the  BrainGlobe ecosystem. This approach works well when an atlas already exists, but not when a lab wants to build a new atlas from scratch. 

That was exactly the challenge faced by researchers at the SWC and the Carl von Ossietzky University of Oldenburg. The Oldenburg group specialises in understanding how migratory birds use the Earth’s magnetic field to navigate long distances — a field in which the Eurasian blackcap is a key model species. However, no detailed blackcap brain atlas existed to support their work. 

Last year, the BrainGlobe core development team enabled producing the very first high‑resolution, 3D digital atlas of the Eurasian blackcap brain. This success sparked interest from other groups who wanted to work with the BrainGlobe team to build their own atlases from scratch. But it also exposed a major barrier: atlas building is complex and very time consuming. 

Our work on Template Builder 

To help the BrainGlobe team meet growing demand, ARC got involved. We focused on improving and automating the first stage of atlas construction: generating a high‑quality, standardised 3D reference template. This step is essential, because only once a template exists can researchers annotate regions and build a full atlas. 

“ARC helped transform brain atlas building from a highly specialised, time-intensive process into a more accessible and reproducible workflow.

Simplified preprocessing workflow 

We reorganised the preprocessing workflow that prepares raw microscopy data for template construction into two clear, reusable functions: 

  • standardise: prepares and normalises the raw microscopy data 
  • preprocess: corrects brightness, masks, crops, pads, and outputs data that is ready to use for template building 

This structure is easier for new users to follow. It also makes the codebase more maintainable and testable. 

Improved testing 

We added unit tests and introduced integration tests to ensure the full workflow behaves as expected. This makes it easier for others to contribute without accidentally breaking functionality. 

Metadata logging 

We improved metadata logging so you always have a clear record of the exact Template Builder version and configuration used during preprocessing. This makes the workflow easier to reproduce and removes the burden of manually keeping track of settings. 

Easier installation 

ANTs (Advanced Normalization Tools) is a widely used software suite for image registration and template construction. It’s powerful but difficult to install on Windows. By replacing ANTs‑based preprocessing steps with lighter, cross‑platform alternatives, we’ve made the tool more accessible. 

Better Quality Control 

QC (Quality Control) plots help users verify whether their data looks correct before moving on to template construction. Plots now show masks overlaid on the underlying image and are collected in a single directory. This makes it much easier to spot issues early and gives researchers more confidence in the data they’re feeding into template construction. 

Improvements to the napari interface 

We added several improvements to the brainglobe-template-builder napari plugin, including exporting configs directly from the interface, more flexible down sampling, new mask‑creation options, and improved metadata logging. We also fixed bugs and improved testing coverage to make sure the plugin keeps working as expected. 

Expanded documentation 

We expanded the documentation regarding required inputs and BrainGlobe’s image‑space conventions. These updates help new users avoid common pitfalls and make the workflow easier to adopt. 

Zebra finch brain template 

To ensure the updated workflow performs well on real biological data, we curated and processed zebra finch datasets using the new pipeline. This provided a valuable test case and helped refining the workflow. 

The bigger picture 

Our work has made template building in BrainGlobe more accessible, reproducible, and easier to maintain, leading to two pre‑releases (v0.1.0 and v0.1.1) and laying groundwork for its first full release. 

Although additional documentation is still needed before Template Builder is ready for general use, the developments so far have made it far easier for the BrainGlobe team to help other groups produce new atlases from scratch. Alessandro already used the updated preprocessing workflow to generate the very first zebra finch brain template, and more species‑specific templates are expected to follow soon. 

Contributors 

ARC: Kimberly Meechan and Stella Prins 
SWC NIU: Alessandro Felder and the rest of the BrainGlobe core development team 
SWC Margrie Lab: Simon Weiler 

More information