help
help > RE: Single-subject ROI-to-ROI connectivity?
Jun 18, 2022 07:06 PM | Alfonso Nieto-Castanon - Boston University
RE: Single-subject ROI-to-ROI connectivity?
Hi John,
Comparing a single subject to a group is perfectly fine/possible. Simply define two groups, with one group having a single subject, and run a standard between-group second-level analysis. The statistics in that case work out as one could expect to compare the measurement from that single subject to the distribution of the reference group (similar to manually computing z-scores), so the threshold for significance is quite high (i.e. it has to be unlikely that this subject came from the distribution of the reference group) but it works perfectly fine (we have also made a small modification to the standard residual randomization procedure for non-parametric statistics to make sure that works also without a problem in this single-subject vs. group scenario).
Hope this helps
Alfonso
Originally posted by John Johnson:
Comparing a single subject to a group is perfectly fine/possible. Simply define two groups, with one group having a single subject, and run a standard between-group second-level analysis. The statistics in that case work out as one could expect to compare the measurement from that single subject to the distribution of the reference group (similar to manually computing z-scores), so the threshold for significance is quite high (i.e. it has to be unlikely that this subject came from the distribution of the reference group) but it works perfectly fine (we have also made a small modification to the standard residual randomization procedure for non-parametric statistics to make sure that works also without a problem in this single-subject vs. group scenario).
Hope this helps
Alfonso
Originally posted by John Johnson:
I found a couple of posts from ~10 years ago
saying single-subject connectivity analyses were not possible after
x version of conn.
Is this still the case?
I have a single person with congenital limb absence, and wolud like to compare their connectivity with my 15 intact participants.
If conn is not an option, is there a good alternative, possibly nipype- or BIDS-app-based?
Thanks,
John
Is this still the case?
I have a single person with congenital limb absence, and wolud like to compare their connectivity with my 15 intact participants.
If conn is not an option, is there a good alternative, possibly nipype- or BIDS-app-based?
Thanks,
John
Threaded View
| Title | Author | Date |
|---|---|---|
| John Johnson | Jun 18, 2022 | |
| Alfonso Nieto-Castanon | Jun 18, 2022 | |