Dear Doug,
The "template_coreg.mgz" image is also upside-down and rotated 90 degrees (X). Can you suggest how I might troubleshoot the initial registration?
I can send a logfile if it helps (not sure if I attach the textfile it will pass through your email filter)
I saw a section in "Samseg.py" concerning SAMSEG_LEGACY_REGISTRATION but I'm not sure if this is relevant.
Sincerely,
Randolph Andrews
# =======================================================================================
#
# Preprocessing (reading and masking of data)
#
# =======================================================================================
# Read the image data from disk and crop
# Historically, the template was resaved with a transformed header and the image-to-image transform
# was later extracted by comparing the input image and coregistered template transforms, but shear
# cannot be saved through an ITK image, so a better method is to pass the image-to-image transform matrix
# directly to samseg. If the SAMSEG_LEGACY_REGISTRATION env var is set, this old method is enabled
if os.environ.get('SAMSEG_LEGACY_REGISTRATION') is not None:
print('INFO: using legacy (broken) registration option')
transformedTemplateFileName = os.path.join(self.savePath, 'template_coregistered_legacy.nii')
self.imageBuffers, self.transform, self.voxelSpacing, self.cropping = readCroppedImagesLegacy(self.imageFileNames, transformedTemplateFileName)
else:
self.imageBuffers, self.transform, self.voxelSpacing, self.cropping = readCroppedImages(
self.imageFileNames,
os.path.join(self.atlasDir, 'template.nii'),
self.imageToImageTransformMatrix
)
-----Original Message-----
From: Douglas N. Greve <dgreve@mgh.harvard.edu>
To: freesurfer@nmr.mgh.harvard.edu
Sent: Sun, Oct 3, 2021 8:04 am
Subject: Re: [Freesurfer] SAMSEG output orientation not the same as input images
Can you compare the template_coreg.mgz to the input image; they
should be mostly aligned. If not, then the initial registration
failed.
External Email - Use
Caution
We have recently installed FS-7.1 and have tested a few
“ADNI” images with SAMSEG. The process runs to completion,
but the output (seg.mgz) appears upside-down and rotated about
90 about the X-axis. We are passing seg.mgz through
BrainStorm to create NiFTI images. We do not believe that
this is a BrainStorm issue inasmuch as the same images run
through recon-all (fully separate proc) yield output images in
the same orientation as the original images. I found a thread
from 2012 with a similar output orientation issue, but the I
did not see a response that resolved this.
Any solution or insight would be greatly appreciated
Sincerely,
Randolph Andrews
(PS. listserv item from 2012)
Re: [Freesurfer] integrating FSL and Freesurfer
Chris Filo Gorgolewski Tue, 09 Oct 2012 01:03:57 -0700
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
MailScanner has detected a possible fraud attempt from "secure-web.cisco.com" claiming to be https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer