Hi Chris,
I'll try that out, thank you so much!
Best,
Lisa
Lisa Jeschke
PhD student
[log in to unmask]
+49 351 463-43897
Cognitive and Clinical Neuroscience / Professur für Kognitive und Klinische Neurowissenschaft
TUD Dresden University of Technology
https://tu-dresden.de/mn/psychologie/ifap/kknw
________________________________________
From: Phillips Christophe <[log in to unmask]>
Sent: 14 September 2023 08:00:00
To: Jeschke, Lisa; SPM (Statistical Parametric Mapping)
Subject: RE: Transformation of beta.nii
Hi Lisa,
The SPM.mat files contains lots of information that are needed to perform statistical inference.
Beyond the full filename of betas, you also have those of any contrast (con*.nii) image that would have already calculated plus the mask.nii (which voxels are considered), resMS.nii (residual map ~= noise variance) and RPV.nii ("resel per voxel" ~= roughness ~= 1/smoothness) images, as well as metadata such as estimated smoothness, variance parameters.,..
So all, in all, warping all the images (betas, resMS, mask,...) renaming them, or updating them in the SPM.mat, could work but that's going to be messy.
If you really want to look at your statistical results, estimated in subject space, in MNI-space. A simpler approach could be to
* write down the threshold you want to use, i.e. the t-value corresponding to p<.001 unc. or p<.05 FWEcor,
* warp the spmT*.nii image,
* display the warped spmT map and manually threshold it to only look at the "significant voxels"
HTH,
Chris
________________________________
De : Lisa Jeschke <[log in to unmask]>
Envoyé : mardi 12 septembre 2023 16:16
À : Phillips Christophe <[log in to unmask]>; [log in to unmask] <[log in to unmask]>
Objet : Re: Transformation of beta.nii
Hi Christophe,
thank you very much for your help and advice!
I have one follow up question, related to my original issue: in case I still want to check my results already in mni-space on the single subject level (after warping my betas into mni), I still need to select my SPM.mat. Do I need to change anything on the SPM.mat after I warped the betas or does it just contain "meta" data about my design, hence does not contain specific information related to the betas that might interfer with the new images?
Thank you very much again,
Lisa
________________________________________
From: Phillips Christophe <[log in to unmask]>
Sent: 07 September 2023 14:23:25
To: [log in to unmask]; Jeschke, Lisa
Subject: RE: Transformation of beta.nii
Dear Lisa,
It is not uncommon to perform the 1st level analysis, aka FFX or within-subject GLM, in subject space. The resulting beta or contrasts images can then be warped into MNI space and entered in the 2nd level analysis, aka RFX or between-subject GLM.
For that 2nd level analysis, the SPM.mat files from each individual are not needed, as you summarize your effect of interest of interest with one contrast image per subject and the build a new GLM.
One quick technical note though.
When warping those contrast images be careful with potential resampling issues. Depending on 1/ the source and destination resolution and 2/ interpolation methods, some artefacts can appear within the image and the edge of the masked-out background can become blurry.
Best,
Christophe
________________________________
De : SPM (Statistical Parametric Mapping) <[log in to unmask]> de la part de Lisa Jeschke <[log in to unmask]>
Envoyé : mercredi 6 septembre 2023 11:10
À : [log in to unmask] <[log in to unmask]>
Objet : [SPM] Transformation of beta.nii
Dear all,
I am currently working on a pipeline with data that requires a lot of memory and processing time. We have calculated a transformation matrix of our T1-registered BOLD series (with ANTs) in order to map it into mni-space.
Due to memory and processing time contraints, I specify and estimate my GLM in SPM before I apply the transforms. That means that I create the SPM.mat and beta.niis and would then like to apply my transformation to only the beta.niis and not all of my "pre-model" images.
However, I do not know how to deal with looking at the results after I have applied my transforms onto the beta-images. Technically I can apply my transform and then save the new betas under the same name and paths, but I am not sure - since the SPM.mat is still the same file - whether the SPM.mat still contains some information on the "old" betas that might interfere, and therefore, if and how I might also need to change things in the SPM.mat?
Does anyone have any experience with that? Thank you very much in advance!
Best,
Lisa
Lisa Jeschke
PhD student
[log in to unmask]
+49 351 463-43897
Cognitive and Clinical Neuroscience / Professur für Kognitive und Klinische Neurowissenschaft
Technische Universität Dresden
https://tu-dresden.de/mn/psychologie/ifap/kknw
|