help
help > RE: Denoising destroys preprocessed nifti output
Oct 14, 2019 08:10 PM | Alfonso Nieto-Castanon - Boston University
RE: Denoising destroys preprocessed nifti output
Dear Daniel,
That sounds mostly like an issue with your analysis mask. For reference, there are three masks that are used as part of the denoising step, the 'analysis mask' (defined in Setup.Options and simply used to define which voxels should be included in the analysis), the 'White matter mask', and the 'CSF mask' (both defined in Setup.ROIs, often automatically during any of the "segmentation" preprocessing steps, and used to characterize BOLD noise components as part of aCompCor). In your case, and given the somewhat non-standard nature of your preprocessing pipeline, I would suggest to:
1) change your analysis mask to use the option 'implicit masking' (using your preprocessing steps CONN's default analysis mask -in MNI-space- would not be in the same space as your functional data)
and 2) change the order or your preprocessing steps to 'functional_art', 'functional_coregister_affine',' structural_segment', 'functional_segment' '& 'functional_smooth' (using your original order, the gray/white/csf masks would also not be in the same space as your functional data)
Hope this helps
Alfonso
Originally posted by Daniel van de Velden:
That sounds mostly like an issue with your analysis mask. For reference, there are three masks that are used as part of the denoising step, the 'analysis mask' (defined in Setup.Options and simply used to define which voxels should be included in the analysis), the 'White matter mask', and the 'CSF mask' (both defined in Setup.ROIs, often automatically during any of the "segmentation" preprocessing steps, and used to characterize BOLD noise components as part of aCompCor). In your case, and given the somewhat non-standard nature of your preprocessing pipeline, I would suggest to:
1) change your analysis mask to use the option 'implicit masking' (using your preprocessing steps CONN's default analysis mask -in MNI-space- would not be in the same space as your functional data)
and 2) change the order or your preprocessing steps to 'functional_art', 'functional_coregister_affine',' structural_segment', 'functional_segment' '& 'functional_smooth' (using your original order, the gray/white/csf masks would also not be in the same space as your functional data)
Hope this helps
Alfonso
Originally posted by Daniel van de Velden:
Dear forum,
I am working with the conn_batch system in order to preprocess and denoise my fmri data for further analysis steps.
My preprocessing steps inherit:
- structural_segment
- functional_segment
- functional_art
- functional_coregister_affine
- functional_smooth
After the last smoothing my fmri data looks fine and ready to be denoised.
Here for I use linear detrending and denoising for the standard conpounds: wm, gw, cfs and realignment.
After that i convert the matc file to a nifti file.
However, the ouput nifti of this denoising steps is cut off by around half the brainheight in x-y-plane. The nifti cotains values, but those are equal to the "not in the brain" voxels.
No need to say, that those results are useless for any further analysis.
Can anybody comment on this kind of error? Anyone experienced the same ? Anyone knows how to fix it, since I tried a whole day and night and couldnt find the reason for this mischief....
Greetings and all the best
Daniel
I am working with the conn_batch system in order to preprocess and denoise my fmri data for further analysis steps.
My preprocessing steps inherit:
- structural_segment
- functional_segment
- functional_art
- functional_coregister_affine
- functional_smooth
After the last smoothing my fmri data looks fine and ready to be denoised.
Here for I use linear detrending and denoising for the standard conpounds: wm, gw, cfs and realignment.
After that i convert the matc file to a nifti file.
However, the ouput nifti of this denoising steps is cut off by around half the brainheight in x-y-plane. The nifti cotains values, but those are equal to the "not in the brain" voxels.
No need to say, that those results are useless for any further analysis.
Can anybody comment on this kind of error? Anyone experienced the same ? Anyone knows how to fix it, since I tried a whole day and night and couldnt find the reason for this mischief....
Greetings and all the best
Daniel
Threaded View
| Title | Author | Date |
|---|---|---|
| Daniel van de Velden | Oct 14, 2019 | |
| Alfonso Nieto-Castanon | Oct 14, 2019 | |
| Daniel van de Velden | Oct 16, 2019 | |