External Email - Use Caution
Dear FreeSurfer developers
I am processing longitudinal 3T T1-weighted data (Freesurfer 7.1.1) from the same scanner and protocol at 3 timepoints across childhood. The first two waves processed without major issues, but in the current pre-adolescent wave I am seeing consistent segmentation failures, mainly in occipital regions (sometimes frontal/parietal). Recon-all “misses” these regions, which were also excluded from the brain mask.
I managed to correct the brain masks by fixing skull stripping (using ms_watershed) and rerunning recon-all, but the segmentation issues persisted. Inspection of norm.mgz suggests residual intensity inhomogeneity as posterior white matter intensity is substantially higher than expected (up to ~140–150). The brain mask also appears brighter here. Could this be causing recon-all to “miss” these regions?
From what I’ve read, a potential solution is running N4BiasFieldCorrection on the raw T1 prior to recon-all (then processing from scratch), and only using control points if needed afterward. Would this be an acceptable approach, particularly in a longitudinal developmental dataset? Any alternative suggestions are also welcome!
Many thanks in advance for your advice.
Kind regards, Chanelle
Disclaimer - University of Cape Town This email is subject to UCT policies and email disclaimer published on our website at https://secure-web.cisco.com/1xGpoPcjPQtqGoQemnEYmTxxk01Cml8DAbleqc_se8czt_w... or obtainable from +27 21 650 9111. If this email is not related to the business of UCT, it is sent by the sender in an individual capacity. Please report security incidents or abuse via https://secure-web.cisco.com/12QWH4f3q4K3n4d15-z-1Z4Ye2973t6xEKRh2n4qKT28F1n...
I suspect that the bright WM is causing your problem. N4 is run as part of recon-all v7 and beyond. You might try running v8 as that uses synthstrip, which is much more robust.
On 2/18/2026 7:57 AM, Chanelle Hendrikse wrote:
External Email - Use Caution
Dear FreeSurfer developers
I am processing longitudinal 3T T1-weighted data (Freesurfer 7.1.1) from the same scanner and protocol at 3 timepoints across childhood. The first two waves processed without major issues, but in the current pre-adolescent wave I am seeing consistent segmentation failures, mainly in occipital regions (sometimes frontal/parietal). Recon-all “misses” these regions, which were also excluded from the brain mask.
I managed to correct the brain masks by fixing skull stripping (using ms_watershed) and rerunning recon-all, but the segmentation issues persisted. Inspection of norm.mgz suggests residual intensity inhomogeneity as posterior white matter intensity is substantially higher than expected (up to ~140–150). The brain mask also appears brighter here. Could this be causing recon-all to “miss” these regions?
From what I’ve read, a potential solution is running N4BiasFieldCorrection on the raw T1 prior to recon-all (then processing from scratch), and only using control points if needed afterward.
Would this be an acceptable approach, particularly in a longitudinal developmental dataset? Any alternative suggestions are also welcome!
Many thanks in advance for your advice.
Kind regards, Chanelle
Disclaimer - University of Cape Town This email is subject to UCT policies and email disclaimer published on our website at https://www.uct.ac.za/main/email-disclaimer or obtainable from +27 21 650 9111. If this email is not related to the business of UCT, it is sent by the sender in an individual capacity. Please report security incidents or abuse via https://csirt.uct.ac.za/report-incident
Freesurfer mailing list --freesurfer@nmr.mgh.harvard.edu To unsubscribe send an email tofreesurfer-leave@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman3/lists/freesurfer@nmr.mgh.harvard.e...
freesurfer@nmr.mgh.harvard.edu
-
Chanelle Hendrikse -
Douglas N. Greve