Hi,
This is a tricky issue as breathing causes major changes in CO2 and
hence the BOLD
effect due to vaso-dilation effects. It also causes changes in the B0
field due to changes
of air in the lung cavity. So both of these will be effecting your
FMRI data and obscuring
changes relating to real neuronal-induced activity. You might be able
to reduce the
effects of B0 change if you have physiological recordings of
respiration (e.g. bellows)
and use retroicor or similar techniques. You could potentially also
model changes due
to CO2 if you have recordings of exhaled gases. Without these you
will really struggle
to extract anything meaningful from this data I'm afraid. For more
details on these kind
of effects I can recommend papers by Richard Wise as well as those by
Rasmus Birn.
All the best,
Mark
On 23 Mar 2008, at 16:12, Heather Urry wrote:
> Hello FSLers: I'm writing for some advice on how to handle analyzing
> an event-related fMRI dataset with four conditions, one of which
> explicitly asks subjects to take a deep breath. The other three
> conditions involve various instructional sets without the deep
> breath. (I have psychophysiological data on a separate set of
> subjects indicating that only those in the breathe condition
> actually do take a deep breath.) At the higher level across
> subjects, I observed lower signal in the whole brain for the deep
> breath condition compared to the other three, which is not
> surprising. I then took a second pass at the first level, this time
> turning intensity normalization on, hoping this might correct for
> global signal differences between the breathe condition and all the
> rest. This correction, however, has the effect of producing tons of
> white matter and ventricle activation for the breathe condition
> versus the others in higher level analyses (see http://www.tufts.edu/~hurry01/ebbl/images/rendered_thresh_zstat2.png)
> . This seems to be an artifact of intensity normalization but I'm
> not sure how to proceed. All the comparisons between the remaining
> three conditions look "normal"-- it's only those with the breathe
> condition in which I'm seeing the weirdness. I considered applying a
> gray matter mask but it occurs to me that I'm not sure the gray
> matter activations that remain aren't artifactual as well. Any
> suggestions or ideas? Know of any references that might help? (I
> don't have respiration rate or depth data.) Whereas in many studies
> we would worry about breathing introducing artifactual signal
> change, in this study I'm actually interested in breathing-related
> signal change and how it compares to signal change in the other
> three conditions. Thanks in advance! Cheers, Heather P.S. I posted
> this message once before with a png attached but received a note
> from the listserv indicating that didn't go through. I hope,
> therefore, that I'm not double-posting inadvertently.
|