Problem with resampling

Submitted by Helen Sawaya on Thu, 09/05/2013 - 22:02

 Dear experts

I am running functional connectivity analysis in DPARSFA but the masks are being resampled inaccurately.
I have entered the functional image of one subject in native space (already motion corrected and slice timing corrected). I am clicking the following boxes:

- realign
- nuisance covariate regression (CSF, White matter, global signal)
- smoothing 
- entering user defined brain mask in subject space
- bandpass filtering
- functional connectivity
- extract ROI timecourses
- define ROI: enter ROI image in subject space.

When the software is resampling the masks from the templates, the AllResampled... output images do not properly overlay the input functional image. So I created a folder and called it 'Masks' in which I put grey, white matter and CSF masks already registered to the subject's native space (from the templates) and named the images "AllResampled_CsfMask..." however the software still doesn't recognize the images.

I would appreciate your help on the issue

Thank you!
Helen

AttachmentSize
Image icon brain_mask.png59.66 KB

Submitted by xjtu2011 on Sat, 09/07/2013 - 14:40

In this order (nuisance covariate regression ahead of the analysis), I guess you might not prepare the anatomical images, i.e. the T1 images.
To run successfully in this order, you need put the functional and anatomical images together.

Submitted by Helen Sawaya on Mon, 09/09/2013 - 21:00

 Thank you for your reply.

I included the anatomical image in a directory called T1Img and the functional image in another directory called FunImg. Do you mean that I should include both images in the same directory? Doing so gives an error and the analysis does not run.
I am wondering whether the masks used to regress out the nuisance signals (example AllResampled_CsfMask...) are the ones registered into native space from the masks in the /template directory (example /Templates/BrainMask_07_91x109x91)? Or rather it's the T1 image that is used? How could I use the templates in DPARSF to correctly overlay the subject's functional image and accurately use the resulting white matter, Csf and global signal as nuisance covariates? I attached the participant's functional image and an image of the AllResampled_BrainMask... resulting from the resampling analysis. As you can see it does not properly overlay the participant's functional image.

Please do advise on how I can carry out the nuisance regression, using the masks in the /Templates directory.

Thank you for your help.

Helen

Submitted by xjtu2011 on Tue, 09/10/2013 - 20:59

 “I included the anatomical image in a directory called T1Img and the functional image in another directory called FunImg. ”  It's right to do so, I think.
My email is hmm4181@gmail.com, 

 
 

Submitted by ZHANG_RESTadmin on Wed, 09/11/2013 - 16:09

 if you want to remove the nuisance signals from individual wm, csf, you have to make the indivdiual structure map normalized. So you may need to check the  "nuisance  regression" option below the normalization/segmentation.   and then check "warp mask into individual space"

Submitted by ZHANG_RESTadmin on Thu, 09/12/2013 - 10:26

 and you have to make sure taht the anatomical image was coregistrered to your functional images? making sure of it is very important or else your two types of iamges could not be correctly overlapped ..

Submitted by Helen Sawaya on Thu, 09/12/2013 - 15:03

 Thank you for your reply.

I was able to do the analysis without including the T1 images. Would you please confirm if it is accurate?

I included the participants' functional images already registered to standard space. I clicked on realign, smooth and default mask, and after that I clicked on filter and nuisance covariates regression ( I did not include segmentation or normalization). and finally functional connectivity and extract ROI timecourses.
The analysis appeared to work fine. Would CSF and white matter signal be accurately regressed in the following analysis? (Is it the template masks that were used?)

Thank you
Helen