Dear all,
I am currently analyzing a dataset (task-based) for a gPPI analysis based on the Brainnetome atlas.
In general, I have 6 conditions (3 - control conditions, and 3-treatment conditions), and for each condition pair, I have selected seeds based on my contrasts of the same condition pair (e.g. wtp_control_green vs wtp_treatment_green).
One region that shows significant activation difference is dlPFC, which has three different subregions based on the brainnetome. So, I wanted to combine all of the regions into one, so I created a mask and uploaded it with SETUP -> ROI -> (atlas, which is how andy's brain book recommends to do it). I did the denoising again, looked at my first level analysis, and saw that my mask was seperated into three different clusters (Can you explain why it did that?).
And also, if I go to the second level analysis, is it correct to do a F-Test with that?
Also, I have noticed the literature is not consistent about how to report results (with FWE cluster correction and the p-threshhold). I know that the standard is cluster correction 0.05, and a p-threshold of 0.0001, however, i also find p>0.005 for example. Are you aware of any literature, that is discussing this topic?
Thank you for your help in advance.
Best,
Negin
Dear Negin
Unchecking the option 'atlas file' in the Setup.ROIs tab and re-running the Setup step will stop CONN from trying to break up that ROI file into anatomically separate subsets.
That said, it is also perfectly fine to simply select all three ROIs in the second-level analysis window and performing an F-test across those seeds (e.g. leaving the default 'any effects (F-test)' option in the 'between-sources contrast' field)
And regarding thresholding options, I am clearly biased but I would suggest the following to get started: Nieto-Castanon, A. (2020). Cluster-level inferences. In Handbook of functional connectivity Magnetic Resonance Imaging methods in CONN (pp. 83–104). Hilbert Press. doi:10.56441/hilbertpress.2207.6603
Best
Alfonso
Originally posted by nnjavaheri:
Dear all,
I am currently analyzing a dataset (task-based) for a gPPI analysis based on the Brainnetome atlas.
In general, I have 6 conditions (3 - control conditions, and 3-treatment conditions), and for each condition pair, I have selected seeds based on my contrasts of the same condition pair (e.g. wtp_control_green vs wtp_treatment_green).
One region that shows significant activation difference is dlPFC, which has three different subregions based on the brainnetome. So, I wanted to combine all of the regions into one, so I created a mask and uploaded it with SETUP -> ROI -> (atlas, which is how andy's brain book recommends to do it). I did the denoising again, looked at my first level analysis, and saw that my mask was seperated into three different clusters (Can you explain why it did that?).
And also, if I go to the second level analysis, is it correct to do a F-Test with that?
Also, I have noticed the literature is not consistent about how to report results (with FWE cluster correction and the p-threshhold). I know that the standard is cluster correction 0.05, and a p-threshold of 0.0001, however, i also find p>0.005 for example. Are you aware of any literature, that is discussing this topic?
Thank you for your help in advance.
Best,
Negin