Print

Print

Hi Estelle,

As Steve pointed out, you may need to generate these frequency maps at different individual thresholds (particularly useful for exploratory analyses; e.g. see Figure 3B in http://www.ncbi.nlm.nih.gov/pubmed/17538950). Sometimes it is useful to apply a threshold on the number of subjects of your frequency map (e.g. to keep only voxels that are observed in at least 10% or 25% of your subjects as the frequency map is likely to be dominated by voxels observed in one or two subjects, in particular when using low statistical thersholds; e.g. see Figure 7B in http://www.ncbi.nlm.nih.gov/pubmed/17478103).

Optional: if you need to take into account anatomical variability at the voxel scale (after normalisation), you could consider the voxel itself plus its 18 (or more) nearest neighbours (this can easily be done using a morphological dilation of your binary maps, see for help "spm_dilate" function). 

I hope this helps,

Mohamed



Steve Fleming wrote:
[log in to unmask]" type="cite">Hi Estelle,

It depends on what you consider "activate". The easiest way to do what you are asking is to create a binary map of each subject's activations, by thresholding say at P < 0.001 uncorrected. Save this image, then using imcalc binarise it (i1 > 0). So then you have a stack of binary T-maps, one per subject (consisting of 1s for voxels that pass your threshold, and 0s for those that do not). If you then sum all these images in imcalc (assuming they have been normalised), then the resulting voxel values will equal the number of subjects who activate that voxel. Note however that the numbers are dependent on the threshold you choose.

You can then display these values overlaid onto an averaged structrural in an image viewer (e.g. MRIcron) with an appropriate colour scale.

Hope that helps

Steve

On 1 April 2010 10:32, estelle.raffin <[log in to unmask]> wrote:


 


Dear SPMers,

I would like to create a kind of "frequency maps", showing for every voxels, how many subjects activate this voxel for a given task, with a colour-code for each number of subjects. Does anyone could help me please?

Thanks a lot

Estelle







Estelle RAFFIN, Doctorante,
Service de Medecine Physique et Readaptation
Hopital Bellevue, CHU de Saint Etienne, 42055 Saint Etienne Cedex 02
Tel: 04 77 12 04 67
Email:[log in to unmask]
[log in to unmask]



Pieddemail