GDC for Siemens Syngo XA60
Yash Patel
- Has anyone else encountered this with Syngo XA60 or similar scanner software?
- Is there an existing solution or recommended approach for handling pre-corrected images?
- Any concerns with selectively skipping GDC in PreFreeSurfer/Topup while keeping it for fMRI volume processing?
Glasser, Matthew
You should be able to configure all of that in your MRI protocol. Gradient distortion correction doesn’t cause huge changes, but you do want it to be consistent across scans. We often for research studies
have it all set to off and use the HCP Pipelines. For clinical studies we often have it all on.
Matt.
From:
'Yash Patel' via HCP-Users <hcp-...@humanconnectome.org>
Reply-To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
Date: Wednesday, January 28, 2026 at 1:59 PM
To: HCP-Users <hcp-...@humanconnectome.org>
Subject: [hcp-users] GDC for Siemens Syngo XA60
Hello HCP experts,
We recently upgraded our Siemens Prisma scanner to Syngo XA60 software and noticed that gradient distortion correction (GDC) is automatically applied to T1w, T2w, and Spin Echo field map images (ImageType: "DIS2D"
or "DIS3D"), but NOT to fMRI EPI timeseries (ImageTypeText: "ND").
The Issue: HCP pipelines use a single --gdcoeffs parameter for all modalities. Setting --gdcoeffs=NONE avoids
double-correction of T1/T2/SpinEcho but also skips GDC for fMRI EPI (which needs it).
Questions:
1. Has anyone else encountered this with Syngo XA60 or similar scanner software?
2. Is there an existing solution or recommended approach for handling pre-corrected images?
3. Any concerns with selectively skipping GDC in PreFreeSurfer/Topup while keeping it for fMRI volume processing?
I've identified that GDC is called separately in PreFreeSurferPipeline.sh, TopupPreprocessingAll.sh, and GenericfMRIVolumeProcessingPipeline.sh, so I can branch/modify the scripts to conditionally skip GDC for
T1/T2/SpinEcho when DIS3D is detected, but wanted to check if others have already solved this or have better approaches.
Note: Disabling GDC at the scanner level also disables Intensity Normalization which we want to keep.
Thanks for any insights!
Regards,
Yash
--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
hcp-users+...@humanconnectome.org.
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/c98157aa-7863-4daf-aad1-98f92db658d4n%40humanconnectome.org.
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
Harms, Michael
Yes, this is known change in the behavior (particularly, the allowed options) in various Siemens product sequences, starting with XA more generally.
Hopefully, in the scans in which you used the on-scanner correction, you made sure to specify the 3D (rather than the 2D) version?
IIUC, part of the issue here is that if you want to use Prescan Normalize, you can't turn off GDC on the SEFMs?? (If so, is that true for the CMRR MB sequences, or just in SEFMs implemented with the product Siemens SE sequence?)
No, we have not implemented a work-around for your specific situation (i.e., GDC already applied on the scanner for the SEFMs, but not the BOLDs).
Matt would have to comment on whether he is willing to accept a PR that implements functionality to not apply GDC to the SEFM processing (because already done on the scanner), but still allows one to apply it to the BOLD (because not done on the scanner).
Conceptually, seems mostly like just the addition of another flag, and some minor code changes to make use of that flag.
Cheers,
-MH
--
Michael Harms, Ph.D.
-----------------------------------------------------------
Professor of Psychiatry
Washington University School of Medicine
Department of Psychiatry, Box 8134
660 South Euclid Ave.
St. Louis, MO 63110
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/C9A41545-FA5A-48FD-BD13-DD80211080B3%40wustl.edu.
Glasser, Matthew
I’m happy to accept tested PRs that help users use the HCP Pipelines with emerging data issues.
I’ll note that if you want to use on-scanner static field correction you currently cannot use 3D gradient distortion correction (or you will get severe artifacts). This is a rare usecase in research, but is a clinically important usecase.
Matt.
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/99D8B067-63E8-4835-8CC4-793B76AD77D4%40wustl.edu.
Yash Patel
Yes, we made sure to choose 3D Distortion Correction.
We can turn off GDC for SEFMs and EPIs, just not the T1w/T2w images. It seems like we have 3 variants of data.
1. GDC on without prescan normalization
2. GDC on with prescan normalization
I have worked on and dry tested individual GDCing; by the end of this week, I will have a confirmed subject test run. I will then submit a pull request with this functionality.
Yash Patel
Harms, Michael
Hi,
I’m not following why exactly you are having an issue with the EPI scans (BOLD and SEFMs).
First, which sequences are you using for both of those? Siemens product sequences, or the CMRR sequences?
So, if you have the Distortion Correction filter set to Off in the BOLD/SEFM scans, you're saying that the UI doesn't then give you the option to turn On the Prescan Normalize filter?
--
Michael Harms, Ph.D.
-----------------------------------------------------------
Professor of Psychiatry
Washington University School of Medicine
Department of Psychiatry, Box 8134
660 South Euclid Ave.
St. Louis, MO 63110
From: 'Yash Patel' via HCP-Users <hcp-...@humanconnectome.org>
Reply-To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
Date: Monday, February 2, 2026 at 3:26 PM
To: HCP-Users <hcp-...@humanconnectome.org>
Cc: "Harms, Michael" <mha...@wustl.edu>
Glasser, Matthew
I can confirm that MPRAGE and SPACE are only offering 2D and 3D distortion correction with no “Off” option in XA60 3.0T XR (Prisma).
I think the user is just trying to have parity between the different scans in their acquisition (fMRI, SEFM, T1w, T2w) as we would recommend. This may cause an issue for AABC…
Matt.
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/1ECE1646-B7C0-45AB-BDBF-06A8599B2D16%40wustl.edu.
Harms, Michael
Yes, there is no "Off" setting for the GDC filter for structurals acquired with Siemens product sequences on XA.
But, is the option to select "Save unfiltered" still available (and thereby generate two different recons)? (It should be available; I know it is on XA30 at least).
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/2B42B6D7-DA72-4A2C-A6AA-AFE8D2EAB941%40wustl.edu.
Glasser, Matthew
There is a button under each of GDC and Pre-scan normalize. The GDC button seems to be checkable, but not the Pre-scan normalize button. I think the user wants to have GDC off and PSN on.
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/4DAA138D-9B45-46B2-819E-3944A9B008E4%40wustl.edu.
Harms, Michael
Stepping back, the issue as originally raised (1/28) was not related to structurals, but rather that they had SEFMs with GDC applied at the scanner, while the corresponding BOLD did not. My posts have been related to trying to understand how that came about. @Yash: Was that just an oversight in the setting up of the protocol?
Then with the 2/2 post from Yash, the conversation seemed to shift to be more about the structurals, which are completely different regarding the GDC-related constraints that exist in the product sequences.
So, it has been a bit hard to follow what is fundamental issue that we are discussing here…
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/44A02103-BC98-48F2-88AB-AAD06C9B4CC7%40wustl.edu.
Yash Patel
BOLD: CMRR multiband EPI (cmrr_mbep2d_bold)
SEFMs: CMRR multiband spin echo (cmrr_mbep2d_se)
So the behavior we’re describing is for the CMRR MB sequences, not Siemens product sequences.
2. What we can and can’t do at the scanner
SEFM and EPI: We can turn off GDC for both. So we’re not stuck with GDC-on for SEFMs; we can close GDC for SEFM and EPI if we want.
T1w and T2w: We could not turn off GDC at acquisition on XA60 (no “Off” for MPRAGE/SPACE in our setup).
3. Workaround for T1w/T2w (ND + Prescan Norm)
Because we wanted a “non–gradient-corrected” (ND) T1/T2 that still had Prescan Normalize, we used a scanner-side workaround:
In View & Go, we can get a “non corrected” version of the T1w/T2w.
That gives us images that are effectively ND (no on-scanner GDC) and still Prescan Normalized.
The downside is that these are marked resampled and derived, so they are smoother than the original recon. We checked this with wbcommand's volume-estimate-fwhm and confirmed the derived images are smoother. So it’s a trade-off we’re aware of.
All this stems from the fact that while using the previous OS i.e. Syngo MR E11, here at SBMed we had GDC set as off with prescan normalization for original acquisitions and we wanted to keep that consistent but we could come as close to it as have a GDC off prescan normalized derived/interpolated T1w/T2w image. The original issue (Jan 28) was: scanner applies GDC to T1/T2/SEFM but not to BOLD, and the pipeline’s single --gdcoeffs can’t handle that mix.
Later we realized: we can’t turn off GDC for T1/T2 at acquisition, and we found the View & Go derive workaround for T1/T2 to get ND + Prescan Norm (with the smoothness caveat).
So there are two related pieces: (a) pipeline needs selective GDC control (which I've ’implemented with separate flags), and (b) scanner doesn’t allow GDC-off for T1/T2 at acquisition, so we use the derive workaround when we want ND + Prescan Norm for structurals.
Regards,
Yash
Glasser, Matthew
In my opinion the smoothness issue is a much bigger deal than GDC on vs off. That will have a tremendous impact on things like surface reconstruction, cortical thickness, etc. I would go with 3D GDC for your T1w/T2w with pre-scan normalize on. Then just use the appropriate options for your structural and fMRI in processing. You’ll need to patch things for data where SE fieldmaps and fMRI are inconsistent, but having them inconsistent with the T1w and T2w is not a big deal.
Andreas Bartsch
Hi,
yes. I agree with Matt.
> We checked this with wbcommand's volume-estimate-fwhm and confirmed the derived images are smoother.
What is the actual difference?
Can you try RetroRecon to get what you want? It’s results might not be smoother.
Cheers,
Andreas
To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/513F8FCA-ED03-46F4-A226-9EC544DE765C%40wustl.edu.
Harms, Michael
So, to be clear for anyone else following this, or that may later find this thread:
The "scanner" need NOT apply GDC to SEFMs collected with the CMRR SE sequence ('cmrr_mbe2pd_se'). That was error on your part in setting up the protocol, under which you have already unfortunately acquired some data. For that issue, you now need a modified version of the fMRIVolume pipeline that will allow one to separately control whether GDC gets applied in fMRIVolume to the BOLD vs. the SEFMs.
Regarding your "T1w/T2w ND" workaround:
How exactly are you generating that via View & Go?
And regardless, have you tried the alternative of selecting the "Save Unfiltered" option in the Distortion Correction filter of the T1w and T2w scans?
That will *automatically* generate a second recon, which will an "ND" version, but I think that particular option may operate on the k-space data (rather than an existing image volume already in the View & Go), and thereby avoid the smoothness distinction that you are raising. (This is similar to Andreas suggestion to try an actual RetroRecon, but much easier, since "Save Unfiltered" is an option you can permanently save into the stored protocol, and doesn't require a separate manual RetroRecon with every scan)
Cheers,
-MH
--
Michael Harms, Ph.D.
-----------------------------------------------------------
Professor of Psychiatry
Washington University School of Medicine
Department of Psychiatry, Box 8134
660 South Euclid Ave.
St. Louis, MO 63110
Andreas Bartsch
Hi,
yes – Michael is absolutely right.
If you can “save unfiltered”, you should be fine.
If not, RetroRecon >should< work but you need to do it manually.
Cheers,
Andreas
To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/7FA3463C-CB40-497E-A5F0-B168400EE64C%40wustl.edu.