Call for comments on Brain Imaging Data Structure

Hi all,
I’m passing this on from the Brain Imaging Data Structure working group. BIDS is an effort to create a coherent representation of data from many neuroscience modalities such as MRI, EEG and similar - 3D microscopy image is one of their targets. They are now examining similar issues to those being discussed in OME-NGFF and they are considering OME-NGFF as the standard representation for microscopy data. One stumbling point is similar to one topic being discussed in OME-NGFF - how to substantially link different microscopy runs to each other spatially. They are collecting use cases and have the following question which I am passing onto you all:


We are putting together a BIDS extension proposal to accommodate microscopy images. This will help the community share data in a standardized way. More details here:
https://docs.google.com/document/d/1Nlu6QVQrbOQtdtcRarsONbX5SrOubXWBvkV37LRRqrc.

In order to make this new standard as widely applicable as possible, we would like to get a better understanding of use cases in the wider histology community.

In particular, we are keen to understand how different users encode the spatial location of multiple files coming from the same sample. For example, when stitching images together within the XY plane, or creating a Z-stack stack of images in 3D.

The key question is: when collecting several images from the same sample, do you rely on an intrinsic coordinate system from the microscope? If so, how is this coordinate system encoded: is it in the native proprietary file format? In a standard format like OME-TIFF? Or, do you rely on an external metadata file?

If there is no coordinate system, how then do you reassemble images?

Thank you in advance for your input and for helping us create a data standard for our community.


Thanks,
Lee

5 Likes

Hi @Lee_Kamentsky.

Thanks for bringing this up. This is certainly a use-case that has been brought up several times. Regarding the storage of such a coordinate system, I’d say we’re working towards a unified solution, but right now the issue that seems most related and active is:

Unfortunately, for finding this metadata, the answer as so often is: it depends.

Speaking only for OME-TIFF initially, there is not a clear place to store this, which is why it’s so exciting to have one evolving in the NGFF work.

~Josh

1 Like

Hi @Lee_Kamentsky .
Very interesting to see the draft doc and thanks for sharing it. As you may have seen in other postings e.g.: Metadata for OME Next Generation File Format (NGFF) within BioImaging North America, and now as an international initiative under QUAREP-LiMi (WG 7 – Metadata – QUAREP), a similar effort to extend a current model (the OME model in our case) has been underway. There appear to be quite some overlap and some differences but it might be useful for both teams to be aware of each other’s efforts. The current state of our proposed set of extensions to the OME model is here: GitHub - WU-BIMAC/MicroscopyMetadata4DNGuidelines: The 4DN + BINA Tiered System of Microscopy Metadata Guidelines was developed by the 4DN Imaging Standards Working Group in collaboration with the BINA Data Management and Quality Control Working Group (https://www.bioimagingna.org/qc-dm-wg) for documenting microscopy experiments and assessing the quality of image data. and available for comments using the github issue system.

On the specific issue of real-world coordinates, and this is part of what @joshmoore refers to, we are proposing a way to store that in the metadata in the form of defining the origin coordinate system and a 4x4 affine transformation matrix of each image plane in a dataset. Exactly where and how that information will be stored is still under discussion and it would be great if we can come up with a common solution.
Cheers,
Damir

3 Likes