Dear Chris,
> It appears masking is a binary operation-- does this mean the mask
> specified must be in a bitmapped {0,1} format, or just that it is treated
> that way?
The latter. The mask can have any numbers. If the mask image format
(e.g. 'float') supports NaN, NaN is the masking value, otherwise it is
0.
>
> If this is the case, is there any way to modulate/parameterize this-- i.e.,
> if an averaged, cropped brain image is taken as a probability "cloud",
> rather than a go/no-go switch?
>
> If this were possible, I should think it would aid in the detection and
> interpretation of signals at the edges of cortex, where "jitter" from
> subject to subject-- even with the great advances in spatial
> normalization-- is still a problem.
Practically, the safe way is to design a mask image, which includes
everything, where signal might come from and add a bit more, just to be
safe that even some residual anatomical 'jitter' is included.
Alternatively, you could construct a mask image (somehow) for each
normalized subject and compute the logical OR mask for these individual
masks.
However, if there is 'jitter' (for some reason) in some subjects, the
sensitivity is decreased anyway at these voxels, because some subjects
show some signal and some don't.
So, I cannot really see, what a rather 'non-intelligent' explicit
masking can do about this, if there is still some 'jitter' after running
the spatial normalization based on some sophisticated algorithms...
Stefan
--
Stefan Kiebel
Functional Imaging Laboratory
Wellcome Dept. of Cognitive Neurology
12 Queen Square
WC1N 3BG London, UK
Tel.: +44-(0)20-7833-7478
FAX : -7813-1420
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|