help > RE: NBS and BrainNet Viewer
Jul 28, 2018  05:07 PM | Andrew Zalesky
RE: NBS and BrainNet Viewer
Hi Hex_Dex,

Yes - you are correct: a 1 in the contrast vector tests for an increase in pre relative to post. A -1 will test for an increase in post relative to pre.

Rather than taking values greater than your test statistic threshold, I suggest taking the edges that were identified as significant. The nbs variable will store a matrix in which significant edges are marked with a 1, and all other edges are marked with a zero. You can use this to mask out the relevant edges. 

You could either present the mean value or the t-statistic value for each edge. I think both are reasonable measures.

Originally posted by hex_dex:
Hi Andrew and everyone,

I want to test the size of effect of an intervention, so I check the main effect of time since I have a repeated measures experiment (in a pre and post time). In my design matrix I put 1s for the pre and -1s for the post and when I run the t test through NBS I get results for both 1 and -1 vector. If I get it right, the 1 significance shows greater value on the pre measurement and respectively, the -1 shows greater value on the post measurement. So I want to depict these changes onto BrainNet Viewer. In order to do that I extract the nbs.NBS.test_stat and only take the connections with a greater value than my threshold value so that I can isolate the connections that do have a statistic significance. My question is: in order to depict that onto BrainNet Viewer, should I take the (masked) connections of all the datasets that I have and calculate their mean value? Is that valid? If not, how can I depict those results?

Thanks a priori,


Threaded View

hex_dex Jul 28, 2018
RE: NBS and BrainNet Viewer
Andrew Zalesky Jul 28, 2018
hex_dex Jul 28, 2018