help > Discrepancy between results viewer in Conn, and SPM 12, within-network connectivity
Showing 1-2 of 2 posts
Mar 27, 2020 01:03 PM | Natasha Mason
Discrepancy between results viewer in Conn, and SPM 12, within-network connectivity
Hello all,
I recently did an ICA analysis, comparing the effects of drug vs placebo on various resting state networks. When I load the results viewer, I see the hypothesized difference between my treatment (reduced within-network connectivity), however I also see some clusters that demonstrate increased connectivity. On further inspection, I see that these clusters of increased connectivity actually lay outside my "core" network of interest, and thus are not of interest to me.
What I have tried to do is overlay a mask of the network, so that I am only looking in the "core". I could not figure out how to do this within conn, so instead I displayed my results in SPM display (via the conn results explorer), with my mask overlaid.
When I do this, my results make more sense (in that I no longer have this increased connectivity with an area outside of the network). However, I see different statistics in SPM (specifically, larger clusters, and they are more significant). I am wondering why there is this discrepancy.
**I realize I had overlaid a mask, and thus this would be changing the statistics, so I also viewed the results in SPM without the mask, and still see a discrepancy between conn results viewer and spm. Again, in SPM the clusters are bigger and more significant.
Best,
Natasha
I recently did an ICA analysis, comparing the effects of drug vs placebo on various resting state networks. When I load the results viewer, I see the hypothesized difference between my treatment (reduced within-network connectivity), however I also see some clusters that demonstrate increased connectivity. On further inspection, I see that these clusters of increased connectivity actually lay outside my "core" network of interest, and thus are not of interest to me.
What I have tried to do is overlay a mask of the network, so that I am only looking in the "core". I could not figure out how to do this within conn, so instead I displayed my results in SPM display (via the conn results explorer), with my mask overlaid.
When I do this, my results make more sense (in that I no longer have this increased connectivity with an area outside of the network). However, I see different statistics in SPM (specifically, larger clusters, and they are more significant). I am wondering why there is this discrepancy.
**I realize I had overlaid a mask, and thus this would be changing the statistics, so I also viewed the results in SPM without the mask, and still see a discrepancy between conn results viewer and spm. Again, in SPM the clusters are bigger and more significant.
Best,
Natasha
Mar 27, 2020 01:03 PM | Jeff Browndyke
RE: Discrepancy between results viewer in Conn, and SPM 12, within-network connectivity
The CONN results are being displayed as two-tailed significance;
whereas, in SPM the results are always one-tailed directional.
This probably accounts for the difference. Try doubling
the p-value threshold in CONN w/ two-tailed significance and see
your negative (or positive) cluster stats now match what is in SPM
output.
Hope this helps,
Jeff
Hope this helps,
Jeff