 |
 |
Hi Alberto, > are there any other rules of thumb? for example, do I have to put many other neurosynth masks, or it's completely up to me determine how many and which RSN include? It's completely up to you how many/what masks to include. It is useful to think of each 3D input to dual regression as a regressor in a GLM design matrix, because that's basically what they are. Each 3D input will explain a portion of the data. This is why, when using the components generated by group melodic as the input to dual regression, it is advised to use all of the components, because this is what will best explain the data. If you leave a component out, the part of the data that would otherwise be explained by that component will be incorrectly assigned to the other components, or will be explained as noise. When you are using inputs derived from a different data set, it is a bit more difficult to ensure that they will fully explain the data, but the same concept applies. > is the dual regression clever enough to prevent false positive After the two regressions have been performed, statistics are generated for every 3D input. You can simply look at the results for the masks that you are actually interested in, and ignore the rest. It's exactly the same as a task-based GLM - you have variables (masks) of interest, and other confound variables that you don't care about. Cheers, Paul On 17/10/2018, Alberto Inuggi <[log in to unmask]> wrote: > hi paul. > thanks for your answer. > are there any other rules of thumb? for example, do I have to put many other > neurosynth masks, or it's completely up to me determine how many and which > RSN include? > is the dual regression clever enough to prevent false positive, i > mean...connectivity associated to the few masks used, but that would not fit > at all in case many others masks has been included. > I mean, is clever enough to say the 90% of the bold activity I see in users' > data does not have anything to do with these (e.g.) 2 given masks? or the > algorithm would end associating "lot of" bold activity to what I gave him > > thanks in advance > Alberto > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1