Integration of BIDSme into Neurodeks env

[SPITZ-Bradley]

Hello Neurodesk team,

I would like to suggest the integration of BIDSme, a neuroimaging tool developed at the University of Liege Cyclotron Research Center notably by BELIY Nikita (GitHub profile). BIDSme is an open-source python tool that converts source-level (raw) neuroimaging datasets into BIDS-compliant formats.

Here is the official BIDSme GitHub repository :
https://github.com/CyclotronResearchCentre/bidsme

I'm currently undertaking an Internship under the supervision of PHILIPS Christophe (GitHub profile), GIGA CRC In vivo Imaging's research director, during which i have worked on the Dockerization of BIDSme :
https://github.com/CyclotronResearchCentre/BIDSme_containerisation

This repo contains all necessary Docker files and user documentation to build and run BIDSme as a containerized application.

I would like to explore the possiblity of integrating BIDSme into the Neurodesk environment to provide users with an efficient, reproducible workflow for BIDS dataset preparation.

Please let me know if I an provide any further details or assistance regarding this integration.

Thank you very much for your attention,
Best regards,
Bradley.

[stebo85]

Dear @SPITZ-Bradley

That would be great :) can you add it with our container builder tool: https://www.neurodesk.org/neurocontainers-ui/

You can simply use your container as the base image, add documentation and the name of the binary and ideally a little test :)

the rest is automated from there

[stebo85]

Dear @SPITZ-Bradley,

This is now included in Neurodesk. Can you confirm it's all working as expected?

[SPITZ-Bradley]

Dear @stebo85, yes, the application works as expected, thank you for your help !

[SPITZ-Bradley]

Dear @stebo85, I have two questions that came up while testing (and sometimes torturing) my BIDSme applications in Neurodesk:

1) Would it be possible to import BIDSme as a Python module (like Nipype) instead of only using the CLI?

2) How does Neurodesk’s JupyterLab interact with loaded modules?
Specifically, BIDSme relies on optional external modules (e.g., dicom2niix, mne).
I’d like to know if there’s a way to avoid installing these optional modules inside the BIDSme container, and instead use the versions that are already available in Neurodesk through other loaded containers.

I tried to find information on these topics but didn’t have much success. Thanks in advance for your feedback!

[stebo85]

1. we are currently working on a feature to connect to a kernel inside a container - but this is not ready yet. The best workaround for now would be to install bidsme via pip in the main Jupyter environment (like nipype). Then it can access all other containers as well after module loading them (see our examples).

2. if you want your current bidsme inside a container to access mne or dcm2niix then the best way would be to install these inside the bidsme container

[SPITZ-Bradley]

Thank you for the clarification and the quick replies.

Actually, I was wondering if it’s possible to install BIDSme directly in Neurodesk inside a virtual environment instead of using a container. Since BIDSme is pure Python, containerizing it may not be strictly necessary.