help
help > RE: Siemens Prescan Normalization
Sep 27, 2016 08:09 PM | Alfonso Nieto-Castanon - Boston University
RE: Siemens Prescan Normalization
Hi Bob,
That is an interesting question. As far as I can tell (I do not really know the details) prescan normalization seems to apply a voxel-specific scaling factor to commpensate for field heterogeneity. If that is correct then you are probably right that in terms of correlation measures, those scaling factors may seem irrelevant. Yet there are several places along the processing pipeline where this sort of between-voxel scaling factors may come into play. Just to name a few that come to mind: a) during realignment and coregistration, these factors may influence the similarity cost function and result in different estimated motion/coregistration parameters; b) these factor may interact with motion to produce motion by heterogeneity interaction effects that may be present in your functional data after realignment (note that this effect is likely not corrected by unwarping, since that deals with spatial-distortion but not scaling heterogeneities); and c) when estimating seed timeseries, particularly for large ROIs, these factor may influence the relative weighting of different voxels and produce different "average" BOLD timeseries. So, overall, I would be wary of using that data directly, but there may be ways to perhaps apply a similar weighting factor post-acquisition (perhaps other people more familiar with the effect of prescan normalization could chime in with some thoughts/comments/suggestions)?
Hope this helps
Alfonso
Originally posted by Bob Kraft:
That is an interesting question. As far as I can tell (I do not really know the details) prescan normalization seems to apply a voxel-specific scaling factor to commpensate for field heterogeneity. If that is correct then you are probably right that in terms of correlation measures, those scaling factors may seem irrelevant. Yet there are several places along the processing pipeline where this sort of between-voxel scaling factors may come into play. Just to name a few that come to mind: a) during realignment and coregistration, these factors may influence the similarity cost function and result in different estimated motion/coregistration parameters; b) these factor may interact with motion to produce motion by heterogeneity interaction effects that may be present in your functional data after realignment (note that this effect is likely not corrected by unwarping, since that deals with spatial-distortion but not scaling heterogeneities); and c) when estimating seed timeseries, particularly for large ROIs, these factor may influence the relative weighting of different voxels and produce different "average" BOLD timeseries. So, overall, I would be wary of using that data directly, but there may be ways to perhaps apply a similar weighting factor post-acquisition (perhaps other people more familiar with the effect of prescan normalization could chime in with some thoughts/comments/suggestions)?
Hope this helps
Alfonso
Originally posted by Bob Kraft:
We have acquired resting state data on Siemens
Skyra. Unfortunately, Prescan Normalization was turned off. This
has resulted in moderate image intensity shading from the center
out (dark in the center bright at the edges). We have
corrected our protocol but i am wondering what effect (if any) this
shading will have on functional connectivity analysis.
I believe that the effect will be minimal since connectivity is
determined by calculation the correlation of fMRI signal from voxel
to voxel. These seems reasonable to me but I am wondering
what the experts think on this topic.
Thanks for your help in advance.
Bob
Thanks for your help in advance.
Bob
Threaded View
| Title | Author | Date |
|---|---|---|
| Bob Kraft | Sep 26, 2016 | |
| Alfonso Nieto-Castanon | Sep 27, 2016 | |