Cifti-convert gives incomplete maps

[Frederik Bergmann]

Hello,

I created fake niftis from the task activation files with cifti-convert and passed them to SPM. I do that because I could not find a way to do stats in FSL or with the taskbatch script as my task design is a bit tricky.

The files I convert are:

tfMRI_TASK1_Atlas_hp_s2_MSMAll_DeDrift_Resample.dtseries.nii 

I then did stats on the subject and group level in SPM. However, when I back convert the resulting images and visualize the result, it seems quite a lot of information is lost: the maps have holes in them (zeros).

This starts to emerge on the subject level con-maps and then accumulates into the group level t-map as the con maps all mask each other.

I would be thankful for some comments.

I am sure I did something I shouldn't have done.

thanks!

Frederik

[Glasser, Matt]

1. Not sure what “resample” means.
2. What do the images look like?

 

Matt.

--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hcp-users+...@humanconnectome.org.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/379a717d-af54-4b49-b304-bdb09e7f37b6n%40humanconnectome.org.

 

---

The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.

[Frederik Bergmann]

Dear Matt,

1) DeDrift_Resample is the suffix I added to the activation file after running DeDriftAndResamplePipelineBatch.sh.

2) the images look like all gyri have been eroded away and only what was deep in the sulci on the MSMAll inflated brain is left. The only bit that is still intact really is the medial surface and the insula. 

Data is also missing in the volume: the lateral edges of the cerebellum and of the hippocampi are missing.

[Frederik Bergmann]

what is still visible looks nice though and functionally plausible.

is it maybe something regarding the options: -smaller-file and -smaller-dims for cifti-convert?

[Coalson, Timothy Scott (S&T-Student)]

-smaller-file and -smaller-dims just change the dimensions of the fake nifti output file, they don't change the data ordering, that is why they don't have equivalents in the -from-nifti option.

If SPM is attempting to do some spatial steps, like smoothing, then expect very strange things to happen.  I don't know why it would come up with a pattern that is related to cortical folding if you didn't use -volume-to-surface-mapping while attempting to convert back to cifti - to be clear, if you use -cifti-convert to make a fake nifti, you have to use -cifti-convert to retrieve it back to cifti format after you process it, and you must not attempt to resample or align (or smooth, or take the gradient, or...) the fake nifti file.

Tim

---

From: 'Frederik Bergmann' via HCP-Users <hcp-...@humanconnectome.org>

Sent: Tuesday, September 20, 2022 5:17 PM
To: HCP-Users <hcp-...@humanconnectome.org>
Cc: Frederik Bergmann <bergmann...@googlemail.com>; glas...@wustl.edu <glas...@wustl.edu>
Subject: Re: [hcp-users] Cifti-convert gives incomplete maps

 

WARNING: This message has originated from an External Source. This may be a phishing expedition that can result in unauthorized access to our IT System. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/7d13ad74-23c3-4b27-aa69-4af06d33f7f1n%40humanconnectome.org.

[Glasser, Matt]

Post some images.

Matt.

[Harms, Michael]

That’s what I would suspect as well.  Doesn’t SPM typically do some spatial processing? (in which case the “fake nifti” approach doesn’t work).

 

-- 

Michael Harms, Ph.D.

-----------------------------------------------------------

Associate Professor of Psychiatry

Washington University School of Medicine

Department of Psychiatry, Box 8134

660 South Euclid Ave.                        Tel: 314-747-6173

St. Louis, MO  63110                          Email: mha...@wustl.edu

 

From: "Coalson, Timothy Scott (S&T-Student)" <tsc...@mst.edu>
Reply-To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
Date: Tuesday, September 20, 2022 at 5:30 PM
To: HCP-Users <hcp-...@humanconnectome.org>
Cc: Frederik Bergmann <bergmann...@googlemail.com>, "Glasser, Matt" <glas...@wustl.edu>
Subject: Re: [hcp-users] Cifti-convert gives incomplete maps

 

* External Email - Caution *

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/SN6PR01MB499196378699EF30A02291AF814C9%40SN6PR01MB4991.prod.exchangelabs.com.

[Frederik Bergmann]

Dear Matt, Tim, Michael,

Thank you for your input! I found the culprit.

I checked the SPM script I used and realized that the default settings include a 0.8 threshold for voxels to be used for analysis.

matlabbatch{1}.spm.stats.fmri_spec.mthresh = 0.8

I set the default value to zero now.

Now everything worked out fine. The results look flawless.

here some tweets regarding the masking process in SPM:

twitter.com/laurenatlas/status/588763472680321024

best

Frederik

@Matt: sorry I cannot post images from work, as Google Docs is blocked here, which would be necessary to attach pictures!

Will try to remember when back home

[Glasser, Matt]

Great!  Was hoping it was something simple like that.

Matt.