Add different resolution of the BigBrain dataset

[cmadjar]

Purpose

In bigbrain.loris.ca, there are different resolution/registration available for the BigBrain dataset. They will need to be imported in CONP.

Context

There are many big brain datasets available with different resolution and registration technics used. Currently, the BigBrain dataset only contains the 40um 3D blocks.

Possible Implementation

Two possibilities were discussed during the CONP dev call of March 25th, 2020:

1. Add the different versions (aka resolution/registration) as subdatasets of the current BigBrain dataset

2. Create a dataset per resolution/registration

Will probably aim toward solution 1. Later, we will need to make BigBrain a subdataset of the processed dataset.

Related issues

https://github.com/CONP-PCNO/conp-dataset/issues/230

[cmadjar]

List of what is available in bigbrain.loris.ca:

3-D Volumes (MINC and NIfTI files for all)

• Registration MNI ICBM 152 Space

  • Encoding 16-bit gray scale unsigned short
    • 4 volumes of different resolutions (100, 200, 300 & 400um)
  • Encoding 8-bit gray scale unsigned short
    • 4 volumes of different resolutions (100, 200, 300 & 400um)
    • ICBM 2009b Symmetric Template 500um

• Registration MNI ADNI Space

  • Encoding 16-bit gray scale unsigned short
    • 4 volumes of different resolutions (100, 200, 300 & 400um)
  • Encoding 8-bit gray scale unsigned short
    • 4 volumes of different resolutions (100, 200, 300 & 400um)

• Registration Histological Space

  • Encoding 16-bit gray scale unsigned short
    • 5 volumes of different resolutions (100, 200, 300, 400 & 1000um)
  • Encoding 8-bit gray scale unsigned short
    • 5 volumes of different resolutions (100, 200, 300, 400 & 1000um)

3-D Surfaces (Obj and GifTi for all)

• Registration MNI ICBM 152 Space (ICBM 2009b Nonlinear Symmetric)

  • Grey left hemisphere (Obj and GifTi formats)
  • Grey right hemisphere (Obj and GifTi formats)
  • White left hemisphere (Obj and GifTi formats)
  • White right hemisphere (Obj and GifTi formats)

• Histological Space

  • Grey left hemisphere (Obj and GifTi formats)
  • Grey right hemisphere (Obj and GifTi formats)
  • White left hemisphere (Obj and GifTi formats)
  • White right hemisphere (Obj and GifTi formats)

3-D Classified Volumes

• Registration MNI ICBM 152 Space (ICBM 2009b Nonlinear Symmetric)

  • 5 volumes of different resolutions (100, 200, 300, 400 & 1000um)

• Registration Histological Space

  • 5 volumes of different resolutions (100, 200, 300, 400 & 1000um)

[cmadjar]

I listed what is in LORIS when navigating through the LORIS website.

When I looked at the FTP website, I see that there are 2 releases: 2013 and 2015. There appears to be also more dataset types (like ROI, MRI, ...) @glatard any suggestion on what we should do?

[cmadjar]

BigBrain FTP site: ftp://bigbrain.loris.ca/

[glatard]

Hi @cmadjar, I think option 1 is the way to go indeed. We could have 1 dataset with a sub-dataset for each resolution, data type (16/8 bits), and registration. I think surfaces and classified volumes are derived data though, so they could go separately.

@jbpoline and @samirdas may want to chime in here.

[emmetaobrien]

@glatard: that will be straightforward to do and I can proceed whenever the decision is confirmed; still also need to know whether to proceed with the 2013 or 2015 release, or both.

[glatard]

Unless the Khan Lab's dataset used the 2013 release, I would suggest sticking to the 2015 one. Let's consider the decision confirmed by today end of day if there is no objection by then.

[emmetaobrien]

Both of the Khan lab BigBrain datasets we currently have are annotated as from the 2015 release.

[samirdas]

Sounds good to me. :)

On Fri, Mar 27, 2020 at 1:36 PM emmetaobrien notifications@github.com wrote:

Both of the Khan lab BigBrain datasets we currently have are annotated as from the 2015 release.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/CONP-PCNO/conp-dataset/issues/238#issuecomment-605145108, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAD4EYFWKU4DFQBEFE6EEC3RJTPZNANCNFSM4LUJFWCQ .

[emmetaobrien]

I have processed the 3D Volumes and 3D Classified Volumes data, which will be made available as part of https://github.com/CONP-PCNO/conp-dataset/pull/239, and am looking into the 3D Surfaces next.