help
help > RE: Coregistration in default CONN preprocessing
Oct 21, 2021 12:10 PM | jblujus
RE: Coregistration in default CONN preprocessing
Hello experts,
In the case of using the indirect normalization to MNI space pipeline, is it possible to use a T1-weighted fmap for correction or does the fmap need to be T2-weighted for use with the EPI?
Thanks,
Jenna
Originally posted by Alfonso Nieto-Castanon:
In the case of using the indirect normalization to MNI space pipeline, is it possible to use a T1-weighted fmap for correction or does the fmap need to be T2-weighted for use with the EPI?
Thanks,
Jenna
Originally posted by Alfonso Nieto-Castanon:
Dear
Katarzyna,
If you are using the pipeline named "... (direct normalization to MNI space)", that pipeline will perform a separate normalization (non-linear transformation to MNI space) of the structural and functional data, so they both end up in the same space (in MNI), without ever having been explicitly co-registered to each other. This is the recommended approach is you have not explicitly acquired FieldMaps or some other way to compute the spatial inhomogeneities specific to your functional data (mainly because in this case the functional-to-structural transformation is effectively non-linear due to the presence of these spatial distortions).
If you have acquired FieldMaps, that simplifies things since after unwarping the functional-to-structural transformation can be considered linear (effectively a simple rigid-body-transformation), so in this case perhaps you may prefer to use the pipeline named "... (indirect normalization to MNI space)", which will unwarp your functional data, perform an explicit co-registration of your functional to your structural data, then it will normalize (to MNI space) only your structural data and last it apply the same normalization-transformation to your functional data.
Last, if you do not care about MNI space and prefer to have everything in subject-space (e.g. for surface-based analyses), then you may prefer to use the pipeline named "...(in subject-space)", which will simply explicitly co-register your functional to your structural data without any form of additional normalization step. Many other combinations are of course possible, so feel free to use these pipelines simply as a starting point and modify to best fit your needs.
Hope this helps
Alfonso
Originally posted by Katarzyna Siuda-Krzywicka:
If you are using the pipeline named "... (direct normalization to MNI space)", that pipeline will perform a separate normalization (non-linear transformation to MNI space) of the structural and functional data, so they both end up in the same space (in MNI), without ever having been explicitly co-registered to each other. This is the recommended approach is you have not explicitly acquired FieldMaps or some other way to compute the spatial inhomogeneities specific to your functional data (mainly because in this case the functional-to-structural transformation is effectively non-linear due to the presence of these spatial distortions).
If you have acquired FieldMaps, that simplifies things since after unwarping the functional-to-structural transformation can be considered linear (effectively a simple rigid-body-transformation), so in this case perhaps you may prefer to use the pipeline named "... (indirect normalization to MNI space)", which will unwarp your functional data, perform an explicit co-registration of your functional to your structural data, then it will normalize (to MNI space) only your structural data and last it apply the same normalization-transformation to your functional data.
Last, if you do not care about MNI space and prefer to have everything in subject-space (e.g. for surface-based analyses), then you may prefer to use the pipeline named "...(in subject-space)", which will simply explicitly co-register your functional to your structural data without any form of additional normalization step. Many other combinations are of course possible, so feel free to use these pipelines simply as a starting point and modify to best fit your needs.
Hope this helps
Alfonso
Originally posted by Katarzyna Siuda-Krzywicka:
Dear CONN community,
I am a new CONN (v. 17f) user and I must say working with CONN is a great experience, this is indeed a very user-firendly software (which is quite rare in our field..)!
I decided to use the default CONN preprocessing pipeline to prepare my data for analysis. I am however puzzled by the fact that there is no inter-modality coregistration in the pipeline, namely no structural to mean functional coregistration step. I was wondering why is that so, or maybe there is an alternative coregistration step within the procedures already implemented? Is the set center to (0,0,0) for functional and structural an equivalent of coregistration? If that's so, what is the logic behind it?
I tried o find this information in the forum but I haven't found a thread concerning this issue. I would be very gratefull for clarification.
Thank you, Kasia
I am a new CONN (v. 17f) user and I must say working with CONN is a great experience, this is indeed a very user-firendly software (which is quite rare in our field..)!
I decided to use the default CONN preprocessing pipeline to prepare my data for analysis. I am however puzzled by the fact that there is no inter-modality coregistration in the pipeline, namely no structural to mean functional coregistration step. I was wondering why is that so, or maybe there is an alternative coregistration step within the procedures already implemented? Is the set center to (0,0,0) for functional and structural an equivalent of coregistration? If that's so, what is the logic behind it?
I tried o find this information in the forum but I haven't found a thread concerning this issue. I would be very gratefull for clarification.
Thank you, Kasia
Threaded View
| Title | Author | Date |
|---|---|---|
| Kasia Siuda | Jun 28, 2017 | |
| Alfonso Nieto-Castanon | Dec 16, 2017 | |
| Will Mellick | Dec 5, 2023 | |
| jblujus | Oct 21, 2021 | |
| Daniel Berge | Dec 15, 2017 | |