HI again!
Just to make sure I completely understand you...
"Presumably you
want [1 1] to be the *main effect* over time. But to get this using a
linear contrast you need to make sure that the mean effect that comes
from
[1 1] is not driven entirely by either the first or second half of your
design. That is, a big effect in a particular voxel in the second half
of
the experiment but no effect in the first half, may still show up in the
[1
1] contrast (because it's a mean). By masking with *both* [1 0] & [0 1]
guarantees that the effects are present in both the first and second
halves
respectively, at some given threshold."
Thus, if the main effect across the entire task was not being driven by
one half over the other, the [1 1] map masked with [1 0] should
essentially be the same as the [1 1] map masked with [0 1], correct? And
if there are differences, then it suggests that the global main effect
across the task is not a great indicator of what regions are activated
as a function of time. In essence then, [1 1] represents a mean of [1 0]
and [0 1] activation, not a simple addition, correct?
"Not exactly. The contrast [1 -1] will directly give you the voxels
which
were more active in the first than second session. Masking with [1 1]
in
this case isn't necessarily meaningful because it doesn't guarantee that
the voxel is showing a main effect. If you want to limit your [1 -1]
contrast to only those voxels "activated" during both halves of the
task,
then once again you need to make with [1 0] and [0 1]."
So masking [1 -1] with [1 1] is wrong for my purposes because only
voxels commonly active (on average) across both halves will be used in
the mask (assuming my understanding of [1 1] as mean and not summed
activation is correct). If instead, I mask [1 -1] with both [1 0] and [0
1] simultaneously, then it will mask the [1 -1] contrast with I>N voxels
that appear in both halves, correct? And this yields a map of I>N
voxels that were more active in the first half compared to second?
Thanks again for your help, it's much appreciated.
Alex
-----Original Message-----
From: Joe Devlin [mailto:[log in to unmask]]
Sent: Monday, June 07, 2004 5:38 PM
To: [log in to unmask]
Subject: Re: [FSL] FEAT setup
Dear Alex,
>Wouldn't masking [1 1] with [1 0] yield a map of I voxels across the
>entire task that is masked so that only those in the first half appear
>as active? If so, wouldn't this yield the same map as that of the
simple
>(unmasked) [1 0] contrast?
Yes and no. Yes, masking [1 1] would yield only those voxels showing a
main effect of I>N across the whole task which were also present in the
first half but this is not the same as the [1 0] contrast. Presumably
you
want [1 1] to be the *main effect* over time. But to get this using a
linear contrast you need to make sure that the mean effect that comes
from
[1 1] is not driven entirely by either the first or second half of your
design. That is, a big effect in a particular voxel in the second half
of
the experiment but no effect in the first half, may still show up in the
[1
1] contrast (because it's a mean). By masking with *both* [1 0] & [0 1]
guarantees that the effects are present in both the first and second
halves
respectively, at some given threshold.
> From my understanding, If I wanted to find I voxels that were more
>active in the first compared to the second half, wouldn't I mask [1 -1]
>with the [1 1] contrast? I interpret this as asking "of I voxels active
>across the entire task, which are more active in the first compared to
>the second half?"
Not exactly. The contrast [1 -1] will directly give you the voxels
which
were more active in the first than second session. Masking with [1 1]
in
this case isn't necessarily meaningful because it doesn't guarantee that
the voxel is showing a main effect. If you want to limit your [1 -1]
contrast to only those voxels "activated" during both halves of the
task,
then once again you need to make with [1 0] and [0 1].
Joe
Joseph Devlin, Ph. D.
FMRIB, Dept. of Clinical Neurology
University of Oxford
John Radcliffe Hospital
Headley Way, Headington
Oxford OX3 9DU, U.K.
Phone: +44 (0)1865 222 494
Fax: +44 (0)1865 222 717
Email: [log in to unmask]
|