PS In relation to the challenge of defining multiple types of
significant activation, one promising approach is the threshold-free
cluster enhancement implemented in FSL:
Smith SM, Nichols TE (2009) Threshold-free cluster enhancement:
Addressing problems of smoothing, threshold dependence and
localisation in cluster inference. NeuroImage 44:83-98.
Jonathan
On Mon, Feb 13, 2012 at 4:56 PM, MCLAREN, Donald
<[log in to unmask]> wrote:
> The idea is to move towards a method that detect small clusters with
> significant voxels and larger clusters with less significant voxels. I don't
> know RFT well enough to know the implications of testing different
> thresholds at different locations would inflate the the error -- but I'd
> like to know the answer as well.
|