Martin,
The problem is not due to a different acquired resolution, but different reconstruction scripts being used at the scanner, so the underlying DICOM data is actually the same dimensions throughout the study.
This means that some subjects have a bit of squishing in the x/y dimensions. And since it's a non-uniform effect in x/y only, that means the degree of effect depends on the participant's head's orientation in the scanner.
When you say that if it was different processing after scanning it could be fixed, are you suggesting there's something we could do to fix it while still maintaining our original cross-sectional processing and edits, or are you just suggesting that it could be fixed by correcting the dimensions and then re-running cross-sectional processing. We were hoping to avoid the latter as it's several hundred scans, but I suspect that's ultimately what will need to happen.
-Mike
From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces@nmr.mgh.harvard.edu] On Behalf Of Martin Reuter Sent: Monday, April 04, 2016 4:35 PM To: Freesurfer support list Subject: Re: [Freesurfer] Structural resolution and longitudinal processing
Hi Mike,
you can't change this retrospectively. It means that different protocols were used when scanning (or different processing after scanning, before passing it to freesurfer-in that case it could be fixed). It would be good to try to understand what exactly changed between acquisitions (is this on the same scanner/headcoil)? If it is the same scanner and no parameters that affect the image intensity changed, I would not worry too much about this small difference in voxel size. You may also want to check if this happens in all subjects, or only some. And if in some, are they all in one of your groups or distributed across groups, to avoid potential bias.
Best, Martin On 04/04/2016 12:07 PM, Angstadt, Mike wrote: We have a large number of scans that have been processed through the standard cross-sectional stream. Subjects have a number of repeated scans, so we now want to take advantage of the longitudinal stream. When first beginning to run subjects through longitudinal template creation, we noticed a warning message in some cases in the log: ******************************************************************************* WARNING: Image geometries differ across time, maybe due to aquisition changes? This can potentially bias a longitudinal study! Will continue in 10s. ******************************************************************************* Looking into it, there does indeed appear to have been a change in the reconstruction stream that led to images at different times during the study to have slightly different voxel dimensions, changing from 1x1x1.2 to 1.01x1.01x1.2mm voxels. Our question is if there is some way to fix this prior to starting the longitudinal stream but without having to redo the original cross-sectional processing and editing that has been completed? -- Mike Angstadt Research Computer Specialist / PANLab Lab Manager Department of Psychiatry / University of Michigan (734) 936-8229 ********************************************************** Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer