Hi! Dr.YAN
I am doing a degree centrality study with the help of Dpabi.
Significant and desired group difference have been found, if we do not regress out the GS.
Different threshold for connectivity (0.2;0.3;0.4) and scrubbing or not have been used to test the effects of preprocessing options on the results. Group difference are stable.
Yet, GSR(Global Signal Regression) dramaticlly changes those results. We can not find similar results even using the ROI method.
I checked previous similar studies, they usually repilicate results in two circumstance (Regress or not). However, recent studies do emphasize on the meaning of GS and more studies choose to not regress out GS without a additional regression analysis.
My question is that dose it a vital problem if we cannot find similar results when regressing out GS?
Thank You
Bob
Hi Bob,
Hi Bob,
Before reading my answers, please vote us at https://www.missionmainstreetgrants.com/business/detail/7733 (after logging into your facebook).
GSR is a difficult question. Please see http://rfmri.org/content/how-deal-different-results-and-without-global-signal-regression
My suggestion is just go ahead without GSR.
Best,
Chao-Gan
Thanks for such quickly reply
Thanks for such quickly reply.
All the materials you provide really comfort me for a while.
In addition, I read your paper in Neuroimage(1) and found useful suggestions about GSR.
Accorsding to this articel:"GSR was effective in reducing motion-, but not site-related effects;post-hoc approaches removed both effectively.""In light of these points, we recommend using post-hoc standardization approaches over GSR"
If anyone have preprossing problems, you can aslo read this paper
Maybe oneday, I must use it to reply to reviewers.
When I have to submitt this work, I will try psychaitry journals to avoid too many related questions from rs-fMRI experts like you...
(1)Yan, C. G., Craddock, R. C., Zuo, X. N., Zang, Y. F., & Milham, M. P. (2013). Standardizing the intrinsic brain: towards robust measurement of inter-individual variation in 1000 functional connectomes. Neuroimage, 80, 246-262.