help
help > RE: About using weighted GLM in the 1st-level analysis
May 13, 2020 03:05 PM | Alfonso Nieto-Castanon - Boston University
RE: About using weighted GLM in the 1st-level analysis
Yes, the weighted GLM bivariate correlation values are computed as the correlation among the weighted
timeseries, where the weights represent the rectified hrf-convolved
task effects. In particular first the condition vector is convolved
with the canonical hrf and rectified -max 0-, then the original
timeseries are centered by removing/subtracting their weighted-mean
values, then the resulting centered timeseries are multiplied by
their weights, and last the standard Pearson correlation
coefficients are computed from the resulting timeseries.
Hope this helps
Alfonso
Originally posted by ys1j13:
Hope this helps
Alfonso
Originally posted by ys1j13:
Dear experts,
I recently started to use CONN and got a few questions about the analysis in this toolbox.
In particular, I am not sure about how the weighted GLM works in functional connectivity analysis.
Now I'm guessing one possible scenario: the condition vector, which is made from the conditions in the 'SETUP', multiplied with the BOLD timeseries in order to make unrelated BOLD signal to zero.
Is it the real process using weighted GLM in 1st-level analysis?
If it is right, I'm not sure how to get bivariate correlation coefficient using that timeseries. Does the toolbox remove the zero points and calculate the correlation?
Thanks!
I recently started to use CONN and got a few questions about the analysis in this toolbox.
In particular, I am not sure about how the weighted GLM works in functional connectivity analysis.
Now I'm guessing one possible scenario: the condition vector, which is made from the conditions in the 'SETUP', multiplied with the BOLD timeseries in order to make unrelated BOLD signal to zero.
Is it the real process using weighted GLM in 1st-level analysis?
If it is right, I'm not sure how to get bivariate correlation coefficient using that timeseries. Does the toolbox remove the zero points and calculate the correlation?
Thanks!
Threaded View
| Title | Author | Date |
|---|---|---|
| ys1j13 | May 11, 2020 | |
| Alfonso Nieto-Castanon | May 13, 2020 | |