 |
 |
Hi, FLOBS are designed to capture variable heights, dispersions and delays of the HRF, so it makes no sense to add a temporal derivative here. Because your basis functions are combined to produce the fit of the signal you measure you need a F-test to establish to what extent the combination of the basis functions explain significant variance in your data. Thus, you should use the parameter estimates, e.g. at a given coordinate, to reconstruct the modeled HRF (by multiplying with the basis function and summing these up). This will also give you an idea whether what you see is mainly due to height, dispersion or delay or a combination of the three. Hth- Andreas ________________________________________ Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von Stefanie Becker [[log in to unmask]] Gesendet: Donnerstag, 15. September 2011 13:32 An: [log in to unmask] Betreff: [FSL] flobs question -- reformulated Dear FSL experts, I asked some questions a couple of days ago but didn't receive a response. I'll try to make my questions clearer. First, in an FSL course that I attended it was explained that slice timing correction could be done in two ways in FEAT; first, by selecting the correct slice timing correction in the pre-stats tab, or second (the default option), by selecting "Add temporal derivative" in the GLM. I'm using the Optimal/custom basis function to analyse my data at the moment, and there, the option for adding the temporal derivative disappears. So now I'm wondering whether I need to switch on the slice timing correction in the pre-stats tab? Second, I stumbled over a statement from the FEAT 3 Practical, that "One problem with using Basis functions is that we have to use f contrasts to look for a significant effect. These are inherently two-tailed, hence we can not tell the difference between "activation" and "deactivation". " Does this mean that, for the contrast 1 -1, I'd get significant activation on the thresholded Fstat-map for C1&C2&C3 (rendered_thresh_zstat) even if the direction of the effect is -1 1? Or does it mean that I get significant activation for all brain regions that are more active in condition 1 than condition 2, even if both activations are negative relative to a baseline? Third, I've read in a thread that the rendered_thresh_zstat map would also show significant differences if only the shape of the custom basis functions differ between the conditions. Is that correct? If yes, is there something I can do to make sure that my results reflect differences in the percent of BOLD signal change and not just the shape of the function? I'm looking forward to your response. Stefanie.