I would guess that this is a scalefactor (and possibly intercept)
issue, whereby the images are written with some random value in this
field of the NIfTI header. You can check this with:
P=spm_select(1,'nifti','Select the image');
You can set the scalefactor to 1 by:
Nii.dat.scl_slope = 1;
Nii.dat.scl_inter = 0;
On 31 August 2011 17:37, Helen Beaumont
<[log in to unmask]> wrote:
> I've solved this problem with Marko's help, and at his request am copying to the forum.
> Problem - I was trying to segment an M0 image created from a Phillips ParRec file with writeanalyze.m written by Colin Humphries, Feb, 2000, University of California, Irvine.
> The M0 image looks OK when viewed with MRIcro, but if displayed within SPM, while it looks OK, the voxel intensities are ~ 10 ^-43.
> In fact the scaling factor is not set within writeanalyze.m and SPM8 within Windows XP defaults to
> 2 ^ -149. This is probably specific to the Windows environment, or even to the PC spec.
> A temporary solution is to multiply the image by 2^149 with imcalc. If I do this, the segmentation works OK.
> I am sure there is a better solution somewhere.
> Helen Beaumont
> University of Manchester
> Stopford Building, Oxford Rd
> Manchester M13 9PT
> Tel: 0161 275 1259