I'm actually using the Normalise procedure (first estimate then write) of spm to transform anatomical data into the standard MNI space and map them onto the template "standard" colin brain.
I'm using the following estimation options:
source image smoothing --> 8
tempate image smoothing --> 0
affine regularisation --> 'mni'
nonlinear frequency cutoff --> 25
nonlinear iterations --> 16
nonlinear regularisation --> 1
The template used is as mentioned the colin27T1 brain.
As input data I was first using Dicom data, which I transformed with the convert method of spm into analyze format (.hdr and .img). When using those data the spatial processing (transformation) works good. But if I use some other analyze data instead, e.g. exported by BrainVoyager after dooing an inhomogeneity correction with the original Dicom data, the transformation fails. I get some images that are completely distorted and when computing the estimate I get the following warning: Matrix is close to singular or badly scaled. Results may be inaccurate.
Does anybody know what that means?
Both Analyze files are lying differently in space than the colin brain (see attached picture: Analyze from Dicom upper left picture and header, Analyze exported from brainVoyager bottom left picture and header and Colin on the right side), but the both have appropriate header information about the reorientation. I thought SPM is evaluating or including this information in some way, so that it doesn't matter that the data is lying differently in space. But actually it looks like it is of importance. Does anybody know, what kind of image orientation and header information spm, especially the estimate method requires and how to solve this problem?
Verschicken Sie SMS direkt vom Postfach aus - in alle deutschen und viele
auslšndische Netze zum gleichen Preis!