help
help > RE: what is the " normalization" in ALLF or LCOR?
Apr 27, 2022 11:04 PM | Alfonso Nieto-Castanon - Boston University
RE: what is the " normalization" in ALLF or LCOR?
Hi Zaeem,
This normalization options is mostly a subject-normalization step that attempts to equalize the overall distribution (across all voxels) of these measures across different subjects. The rational is similar to cost-based thresholding options in graph theory analyses in that it allows you to focus on relative spatial differences in these measures within each subject while disregarding global (present in all voxels) effects. In general I would probably recommend normalized measures when there is an interest in focusing on relative values within each subject (e.g. where is local correlation highest across different subjects?) or when there is a concern on potential sources of variability across subjects which may affect these measures globally (e.g. motion-induced local correlation differences across subjects), while I would recommend raw measures in all other cases when there is an interest in focusing on absolute values across subjects (e.g. what is the average local correlation in one area across different subjects?)
Hope this helps
Alfonso
Originally posted by Zaeem Hadi:
This normalization options is mostly a subject-normalization step that attempts to equalize the overall distribution (across all voxels) of these measures across different subjects. The rational is similar to cost-based thresholding options in graph theory analyses in that it allows you to focus on relative spatial differences in these measures within each subject while disregarding global (present in all voxels) effects. In general I would probably recommend normalized measures when there is an interest in focusing on relative values within each subject (e.g. where is local correlation highest across different subjects?) or when there is a concern on potential sources of variability across subjects which may affect these measures globally (e.g. motion-induced local correlation differences across subjects), while I would recommend raw measures in all other cases when there is an interest in focusing on absolute values across subjects (e.g. what is the average local correlation in one area across different subjects?)
Hope this helps
Alfonso
Originally posted by Zaeem Hadi:
Hi Dr. Alfonso,
As Emily said above, I was wondering what would be the reason one would prefer using normalization of connectivity values instead of raw connectivity values.
Does normalization improves assumptions for statistical testing?
Would be great to have your opinion on this.
Best,
Zaeem
As Emily said above, I was wondering what would be the reason one would prefer using normalization of connectivity values instead of raw connectivity values.
Does normalization improves assumptions for statistical testing?
Would be great to have your opinion on this.
Best,
Zaeem
Threaded View
| Title | Author | Date |
|---|---|---|
| SUHONG KIM | Dec 22, 2020 | |
| Alfonso Nieto-Castanon | Jan 9, 2021 | |
| Emily Stern | Jul 27, 2021 | |
| Zaeem Hadi | Apr 26, 2022 | |
| Alfonso Nieto-Castanon | Apr 27, 2022 | |