 |
 |
Hi David I simply meant that if you have multiple hypotheses about the cause of the difference in scaling, then in principle, you should include these in models and compare them. In practice, how to do this will depend on your hypotheses ☺ Best Peter From: David Hofmann [mailto:[log in to unmask]] Sent: 06 November 2018 14:09 To: Zeidman, Peter <[log in to unmask]> Cc: spm <[log in to unmask]> Subject: Re: VOI time course shows large differences in magnitude of principal components Hi Peter, thanks for the reply, I have to take a look a closer look at the data. At the moment I do not know what causes these differences. But could you clarify what you mean by modeling the sources? That is, how would I model them in the DCM framework? greetings Am Mo., 5. Nov. 2018 um 15:56 Uhr schrieb Zeidman, Peter <[log in to unmask]<mailto:[log in to unmask]>>: Dear David I don’t think this is an SPM related question… The amplitude of the response will be determined by the physics of the acquisition as well as the generation of the underlying signal. If you look at your functional images, are some areas brighter than others? I.e. are they heavily affected by bias (a slow gradient in signal amplitude)? You can often find the areas closer to the head coil are brighter than deeper areas. You might want to consider increasing the bias correction in your preprocessing if so. As for the effect on DCM… yes I think it might make a difference that some areas have greater scaling than others. All timeseries will be rescaled during DCM estimation (to have maximum variance of 4% if I remember correctly). So the scientific question for you to consider is – what are you hypotheses for the source of this difference in scaling? And do you want to try to model those sources in order to address the question? Best Peter From: David Hofmann [mailto:[log in to unmask]<mailto:[log in to unmask]>] Sent: 24 October 2018 23:09 To: spm <[log in to unmask]<mailto:[log in to unmask]>> Cc: Zeidman, Peter <[log in to unmask]<mailto:[log in to unmask]>> Subject: VOI time course shows large differences in magnitude of principal components Hi all, I used the VOI tool to extract the time course of two different ROIs. Strangely, the magnitude of the principal components are very different between the ROIs (see attached). That is, the first ROI (blue) has a much higher magnitude. Can anybody explain to me how such large differences come about? I also wonder, if this might cause a problem in the estimation of a resting state DCM later on? greetings David