Gradient field inhomogenities and SCT

SCT
1

Dear SCT-team,

We are planning on undertaking a study using T1 brain-MRI scans to analyze the C2-C3 cross-sectional area in longitudinal data. The scans include the brain and upper cervical spinal cord.

The data comes from multiple sites with scanners from different manufacturers (Siemens, Philips, GE). We’ve previously analyzed C2-C3 CSA using brain-scans from Siemens scanners, and prior to running the scans through SCT we processed them with the HCP gradient unwarp script to unwarp gradient field inhomogeneities. C2-C3 CSA was approximately 10-15 mm2 larger in the unwarped data and was more in line with what has been reported in other studies. To our knowledge, the HCP gradunwarp is only possible for Siemens scanners.

Can we use SCT to correct for magnetic field inhomogeneities or does SCT have other features that could help us in this case? We are planning on following this example which I don’t think mentions unwarping explicitly.

If we can’t unwarp the data, would it be recommended to look at relative differences (percentages) over time or would we have too many uncertainties given the gradient field inhomogeneities that we can’t trust the relative numbers?

Sorry for the long post and thank you in advance for your help!

Kind regards,

Carl

2

Hi,

To our knowledge, the HCP gradunwarp is only possible for Siemens scanners.

Looks like it also works for GE, see the code.

Can we use SCT to correct for magnetic field inhomogeneities or does SCT have other features that could help us in this case?

No, SCT is not designed to do that

If we can’t unwarp the data, would it be recommended to look at relative differences (percentages) over time or would we have too many uncertainties given the gradient field inhomogeneities that we can’t trust the relative numbers?

Before trying to unwrap, I would first make sure that the data is not already corrected for gradient non-linearities. You certainly don’t want to apply a deformation field on already corrected data (otherwise you would introduce incorrect deformation).

After you made sure of that, it seems like you can use gradunwarp for Siemens and GE, so that’s a good start. For Philips, I recommend you reach out to a Philips engineer.

Best,
Julien

3

Thank you!