I have been thinking about the scale block, which I have never used. As far as I understand, it should not affect the final statistical group maps of e.g. a task where we compare the beta values. But it would affect the averages of an ROI.
For resting state seed analysis, is it important to use the scale block? We would e.g. use a DMN seed, for each subject take the average time course from the errts time series and preform a regression for each subject on the errts data using the seed time series as a regressor to generate individual correlation maps which we then group compare by via a 3dttest++ using the correlation coefficients.
Would scale on/off make a difference in the final group result? Do you recommend using scale, or is it even a required?
Actually, scaling does have a small effect on the statistical maps. The reason is that scaling is done per run. That should not be a big effect, but it will make the difference more than just truncation size (assuming there is more than just one run). And yes, it would indeed affect ROI averages.
For resting state analysis, with a single voxel as the seed, there would still be that small effect. And many seeds are from radial or ROI averages. While those should not be very important, there would be SOME effect, and therefore there would be some at the group level.
For this type of analysis, scaling is not considered important. Do it or don’t. Our resting state examples currently have scaling included, but just because there are some fringe benefits, depending on what other computations are desired. If all you are running is a correlation analysis, either way should be fine.
rick
The
National Institute of Mental Health (NIMH) is part of the National Institutes of
Health (NIH), a component of the U.S. Department of Health and Human
Services.