Brain regions intensity quantification and visualization

Brainglow has been thought to help quantifying the intensity of brain regions from registered atlas on lightsheet microscopy, and provides visualizations using bar graphs and interactive sunburst diagrams.

Requirements

The required Python libraries are listed in requirements.txt. You can install them using:

pip install -r requirements.txt

Usage

1. Input Files: Prepare the following input files after running brainreg on every channel for your brain (signal and autofluorescence channels):

   • registered_atlas.tiff: Allen atlas labels mask by brainreg, ideally obtained using a reference image like autofluorescence at 561nm
   • downsampled.tiff: Intensity image of your target channel, with high SNR if possible.
   • structures.csv: Region names, IDs, acronyms, and parent structure, from the allen atlas project (included in this repo).
   • volumes.csv: Region volumes, computed by brainreg. Use the one created by the best registration.

2. Intensity Quantification: Run the quantification.ipynb notebook. This notebook loads the input data, calculates the total intensity for each brain region, and saves the results to quanti.csv.

3. Visualization:

   • Use bar_graphs.ipynb to create bar graph visualizations of the quantification results (basic hierarchy support)
   • Use sunburst.ipynb to generate interactive sunburst diagrams for hierarchical exploration of the data.

Output Files

• quanti.csv: Quantification results containing structure name, total intensity, volume, and intensity per volume.
• xxx_sunburst.html : Interactive sunburst charts based on the quanti.csv file to explore the data in html version (can be opened in any browser)
• Bar graphs can be saved easily in png format manually from the jupyter notebook preview if needed.

Hardware

Brainglow uses Dask to parallelize processing tasks and will extensively use your CPU. RAM usage can go up to 30 GB for region intensity calculation (with 10um atlas registration).

Credits

This code has been created in Bellone's lab, UNIGE.