Hi Jeroen,

NB the highres anatomy has no sform or qform (they are empty)

And are the s/qform codes also set to 0? This is the cause of the problem - If this is the case then, when you are using highres as a reference, flirt detects that the reference orientation is unknown, so bases the s/qform of the output image upon that of the source image (the MNI152 in this case).

In the vast majority of cases, using nifti images with unset s/qforms is a very unwise strategy, as it will become very difficult to tell right from left. 

THEN, I open the old fslview_deprecated (3.2.0), here the standard brain overlays nicely with the highres anatomy... weird.

FSLView only uses the s/qform affines to determine left-right orientation, and will overlay any images which have the same number of voxels.  FSLeyes uses the s/qform affines to position images on the display.

Do you have any advice on perhaps handling this issue in a more elegant way?

The solution is to make sure that your highres image has valid s/qform affines, which transform the voxel coordinates into a (roughly) RAS coordinate system before estimating the initial highres->standard transformation.

Cheers,

Paul

On 24 August 2018 at 16:18, Jeroen Siero <[log in to unmask]> wrote:
Hi there,

When using FSLeyes (0.21.1) ,  I encountered something odd when transforming (using inverse matrix of an affine transformation matrix) ROIs in standard MNI space back to highres anatomy space (subject space). The ROIS end up with a huge offset somewhere way outside the brain.

So, I checked all registration steps. Interestingly, I see that the highres anatomy (1mm iso) overlays perfectly on the MNI standard brain (2mm) - the affine transformation matrix should be fine.

However, mapping the standard brain back to the highres anatomy space, same odd thing -> the standard brain is mapped outside the highres anatomy.

THEN, I open the old fslview_deprecated (3.2.0), here the standard brain overlays nicely with the highres anatomy... weird.

I guess it has to do with that FSLeyes perhaps uses the sform/qform information of the header and these are perhaps not copied correctly using inverse transormation with flirt/applyxfm).

After fslcpgeom to copy the highres anatomy geometry info to the standard2highres image, it also looks fine in FSLeyes. You just have to be aware of this but lots of errors might occur. NB the highres anatomy has no sform or qform (they are empty), whereas it seems that the fields are fileld for the standard2highres header...

Do you have any advice on perhaps handling this issue in a more elegant way?

Best,
Jeroen

########################################################################

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1


To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1