New BEP031: Microscopy

[Marie-Helene Bourget]

Hi BIDS Community,

Thanks to many contributors who helped to draft a proposal, we just started a new BEP to extend BIDS to microscopy data (BEP031). We are actively looking for contributors and feedback from the community. You can find the BEP Google Document here: https://bids.neuroimaging.io/bep031

Please feel free to contribute by adding comments and suggestions directly on the google document.

Related discussion thread can be found on GitHub: https://github.com/bids-standard/bids-specification/issues/510. We invite you to keep an eye on it and/or join the BIDS mailing list (https://groups.google.com/forum/#!forum/bids-discussion) to follow the progress of this BEP.

If you are interested (or know someone who would be interested) in joining the core Working Group for this BEP to further discuss and incorporate community feedback, please add your name in the BEP Working Group list at the top of the google document (https://bids.neuroimaging.io/bep031), and make sure to subscribe to the BIDS mailing list.

Best,

Marie-Hélène

[Julien Cohen-Adad]

Dear all, 

I hope this email finds you well.

This email is a follow-up on the initiative described on this thread: the BIDS extension proposal for organizing microscopy data.

We are happy to announce that we now have a solid version of the extension proposal, which is ready to be reviewed: 

https://bids.neuroimaging.io/bep031. 

Please feel free to contribute by adding comments and suggestions directly on the google document (please sign in so we know who is contributing).

In particular, we would like to get your feedback/approbation on the difficult decisions that we had to take:

- Support for both PNG and OME-TIFF file formats

- Support for both 2D and 3D

Lastly: please consider putting your name on the "BEP Working Group" (1st page of the google doc) if you have contributed to this proposal, even if you only left a few comments. Some of the comments were extremely relevant and constructive, so it would be fair to have your name listed as official contributors of this BEP. 

Thank you all very much for your inputs, and special thanks to Marie-Hélène for her amazing work  🎉

Julien

p.s..  …and happy Halloween everyone! 🎃

[Mastaneh Torkamani]

Dear all,

We have recently acquired multiple microscopy collections in VSI format from Olympus slide scanners. Olympus VSI compresses composite files (one label, one overview, and one or more images from actual sections mounted on the slide), and stores them in multiple series, each corresponding to one resolution: 

• filename.vsi (VSI file)
• _filename_ (directory)*
  • stack1 (sub-directory) (or 10000, 10002, 10005, etc)
    • frame_t_0.ets (ETS file)

I can open these images in QuPath/Windows and via Bioformats plugin in Python/Linux. However, converting the largest series to tiff or ome.tiff is time consuming. 

Q1. Is there a toolbox for reading these files and converting them to ome.tiff, png, or other acceptable file formats for BIDS-Microscopy or shall we write one ourselves, and 

Q2. Is there any recommendation for maintaining label and layout info and images for storing them in BIDS-compatible datasets?

Thank you all in advance,

Mastaneh 

mastaneh....@uef.fi 

[Satrajit Ghosh]

dear mastaneh,

the bioformats command line tool may be most helpful instead of doing it interactively. depending on sizes of images, yes it can be time consuming. if you haven't already found this forum, they may be able to offer better suggestions: https://forum.image.sc/

regarding q2 does your use case not fit the chunk configuration options described in figure 1 of the bids-microscopy specification? if it does, i believe the metadata in the json would reflect that info. 

both ome.tif and ome.zarr support pyramidal data. so if you are acquiring data at different resolutions, you may want to store them in a single file. 

cheers,

satra

--
We are all colleagues working together to shape brain imaging for tomorrow, please be respectful, gracious, and patient with your fellow group members.
---
You received this message because you are subscribed to the Google Groups "bids-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/9491c149-c42b-4969-8f38-e6a227fdba52n%40googlegroups.com.

[Mastaneh Torkamani]

Dear Satra,

Thank you for your helpful comments. Indeed, the chunk layout and sequence from figure 1 need to be incorporated for this conversion. I was going through the BIDS-microscopy reference paper to see if a conversion tool was suggested, but it seems this task is still being handled by each group according to their needs. 

I will follow your suggestions as ome-tif seems to be a good solution for saving large data at multiple resolutions. Still, data size and memory issues during visualizations are quite daunting. Other colleagues have suggested storing data in hdf5 containers for intermediate processing, and I'm wondering if the community could consider that as an extension to formats acceptable for the derivatives directory... It seems like it's being discussed anyways https://github.com/bids-standard/bids-specification/issues/197). 

Best,

Mastaneh 

[Satrajit Ghosh]

dear mastaneh,

if scale and visualizations are issues, i would just recommend going directly to ome.zarr. (we tend to use 128^3 + blosc with zstd level 5 compression to optimize storage size and read speed). the ome-zarr py library can be helpful in doing this, but you would need an olympus reader. hdf5 is a good container (and we have stored data using that), but the microscopy community is moving towards zarr for larger scale data for computation and visualization.

cheers,

satra

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/7846fcdc-5f41-420b-aed9-d5ad357d1228n%40googlegroups.com.

[Sébastien Jodogne]

Dear Satra,

You might also find an interest in DICOM for whole-slide imaging, which is an alternative to HDF5 containers, for which free and open-source, reference implementations do exist. For instance in Orthanc:

https://wsi.orthanc-server.com/demo/

https://book.orthanc-server.com/plugins/wsi.html

Kind Regards,

Sébastien-

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CA%2BA4wOmhYjHUTjzW0LdA61vck%2BoD8%2BV-p4CL58bfVW1s8GMZAA%40mail.gmail.com.

--

Sébastien Jodogne
Web: https://www.info.ucl.ac.be/~sjodogne/

Twitter: https://twitter.com/sjodogne

[Dr Cyril, Pernet]

slightly tangential question : is there any matlab support for zarr? bids derivatives recommendation for n-arrays is leaning on zarr and h5 but multiple languages support is important too.

Matlab/Python/R read .h5 files

Pyhton/R read .zarr, I have not seen matlab/octave tool for that (yet)

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/CA%2BRg%2BVzZgTwxhUjNzsc%2B0J5L5BQVO6JKDxOabcfDPQVdsUHw2g%40mail.gmail.com.

-- 
Dr Cyril Pernet, PhD, OHBM fellow, SSI fellow
Neurobiology Research Unit, 
Building 8057, Blegdamsvej 9
Copenhagen University Hospital, Rigshospitalet
DK-2100 Copenhagen, Denmark

wamc...@gmail.com
https://cpernet.github.io/
https://orcid.org/0000-0003-4010-4632