The online BrainPainter service has been migrated to Google Colab. We are not maintaining this Virtual Machine anymore, which is due to be discontinued soon.



BrainPainter

A free software for visualisation of brain structures, biomarkers and associated pathological processes.

How it works


How BrainPainter works





Get Started




Run BrainPainter








Fill-in at least the first two colours.

Colour corresponding to input biomarker value 0

Colour corresponding to input biomarker value 1

Colour corresponding to input biomarker value 2

Colour corresponding to input biomarker value 3

Colour corresponding to input biomarker value 4

If more than two colours are desired for interpolation, make sure the input biomarker range is scaled to 0-5, to use up to the 5th colour for interpolation.



Upload input data file (.csv) according to these templates (max 20 different rows):

Processing usually takes around 15 seconds for each brain (set of 3 images). If you need to process more than 20 images, either send it in batches or do it locally on your computer using the source code.

Please email me (razvan [at] csail.mit.edu) if you encounter any errors.





Example uses of BrainPainter

Cortical - outer Cortical - inner Subcortical
Cortical surface Cortical surface back Subcortical structures

 

Animation created using BrainPainter, showing progression of amyloid deposition in Alzheimer's. By Garbarino and Lorenzi, IPMI, 2019






Brain images generated using BrainPainter, showing atrophy progression in Alzheimer's subtypes. By Young et al., Nature Comms., 2018




Atrophy progression for subcortical regions in Huntington's disease. By
Wijeratne et al, Ann. Clin. Neurol., 2018.




Regions included in the Desikan-Killiany (DK) atlas. By Alexandra Young.





Regions included in the Destrieux atlas. By Alexandra Young.





Regions included in the Tourville atlas. By Alexandra Young.





Regions included in the subcortical atlas. By Alexandra Young.





Advanced Use

For more customisation, please run straight from the source code. No installation is required, as the program is already bundled within a docker container. The following aspects can be customise there:

Generating Colorbar/Legend

For generating a colorbar/legend, you can use this latex script (instructions for using it in the script). It re-creates the default colorbar as on the website, but you can change the exact text labels on the colorbar. To modify/teak it, you will need to copy the project in your overleaf account space.

Q: What if my biomarker data is not in the [0,5] range, but on a different scale/range? If your biomarker data is on a different scale, e.g. [-2, 9.8], you can simply re-scale your biomarker data to 0-5, run BrainPainter to generate the images using [0,5] as scale, then change the labels/tick-marks in the colorbar to [-2, 1, 4, 7, 9.8]. To be very precise, they should be [-2, 0.95, etc ...].

Q: What about using more than 5 colors? You can do that, but need to run BrainPainter from the source code and change the config.py. You can also easily change the colorbar script to add more colors in the legend.

Videos

For generating videos showing e.g. progression of pathology, you need to generate many images (25/sec), where each image is a new row in the input .csv spreadsheet. For example, to generate a 4-second video, you need 100 rows. The numbers you fill in need to change smoothly. After generating the images, you can find programs online/linux commandsto merge the frames into an animated gif.

Finally, please send images in batches of max. 20, as this will otherwise congest our server and the program will also time out. For generating many images (>100), consider running the software on your local machine from the docker container.

Source code

The source code for the BrainPainter software can be found here, while the source code for this website, written in flask, is available here.





News/Updates





About BrainPainter

BrainPainter is a free and open source software published under the CC-BY-SA software license. Our paper, which can be found here here, was accepted for an oral presentation at the MICCAI 2019 MBIA workshop in Shenzhen, China.

Funding for BrainPainter was provided by the NIH grants NIBIB NAC P41EB015902 and NINDS R01NS086905, as well as the EPSRC Centre For Doctoral Training in Medical Imaging with grant EP/L016478/1. We are also particularly grateful to Anderson Winkler for creating the 3D brain models for all three atlases, which are used by the software.



If you experience any issues, please email me (razvan [at] csail.mit.edu) or raise an issue on the github page.