External Email - Use Caution
Dear FreeSurfer Community,
We are currently processing volumetric data from a longitudinal study (adolescent patients with anorexia nervosa vs. healthy controls, 2–3 time points) using FreeSurfer 8.0.0 and the standard longitudinal pipeline (recon-all -long <tpNid> <templateid> -all).
We performed Quality Control (QC) directly on the longitudinal outputs, rating the segmentation for each region individually. Based on this QC, we identified several areas requiring manual correction due to significant overestimation. (We are aware that manual corrections should be reserved for cases with substantial segmentation inaccuracies....)
According to the FreeSurfer documentation, manual edits should ideally be performed on the cross-sectional runs and the base template rather than the longitudinal files themselves. Upon inspecting our data, we observed a few conflicting points:
* Cross-sectional and base accuracy: In many cases, the segmentation in the cross-sectional outputs and the base images actually appears more accurate than in the longitudinal outputs, despite the perceived higher resolution of the latter.
* Lack of "Edit-Trigger": So, in several instances, the specific areas we flagged as problematic in the longitudinal output look fine in both the cross-sectional images and the base template. If there is nothing to "fix" in the earlier stages, re-running the longitudinal pipeline is unlikely to change the erroneous longitudinal output.
This leaves us with a methodological dilemma: Our current QC (identifying errors) is based on the longitudinal output. If we shift to editing cross-sectional images, we would essentially need to redo the QC for those runs, even though they already appear superior to the final product.
What is the recommended procedure in this case? Should we proceed with manual edits directly on the longitudinal outputs despite the documentation’s warnings?
Any insights or recommendations on the best practice for this scenario would be greatly appreciated.
Furthermore, we are questioning whether the application of a longitudinal pipeline is appropriate for our specific study population. Our sample includes adolescents undergoing normative neurodevelopmental processes, alongside patients with Anorexia Nervosa who exhibit significant brain volume reductions (approximately 5%) in the acute stage, followed by substantial recovery at follow-up. We are concerned that these pronounced morphological changes and the resulting high variance could potentially bias the creation of the within-subject base template, possibly affecting the accuracy of the longitudinal registration.
Best regards,
Lara
External Email - Use Caution
Dear Lara,
Even though I am not fully familiar with your pipeline, what you are describing may be related to a longitudinal registration issue, which could be causing the data to appear this way. You may need to correct this problem or consider using a more robust registration method. However, I may be mistaken if your pipeline does not involve any registration steps.
Regards, Ahmed
STATEMENT OF CONFIDENTIALITY: The contents of this email and any attachments are confidential and are intended solely for the addressee. The information may also be legally privileged. This transmission is sent in trust for the sole purpose of delivery to the intended recipient. If you have received this transmission in error, any use, reproduction, or dissemination of this transmission is strictly prohibited. If you are not the intended recipient, please immediately notify the sender by reply email or at 859-693-3221 and delete this message and its attachments, if any.
________________________________ From: Keller, Lara Lara.Keller@lvr.de Sent: Friday, January 23, 2026 9:27 AM To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Manual editing strategy for longitudinal pipeline (FS 8.0.0) following QC
You don't often get email from lara.keller@lvr.de. Learn why this is importanthttps://secure-web.cisco.com/1tCI5XFvF43Gjy4TEQbRWUkCrcsUr6lrpdWOgwcO58U8Tw923l0Sz9M3CepPb-gXg8qPcd-EOI2TgBnWwvQ9jtfCvvpKzwEI6TTWa59jA3p8jFzdXs5ag8oC34ne11fB0ghIsuUYdyEDWPBol7I1hu10-TZXOQ96qWLfHUvQQoGCMz2-LjiTo_sP_7WHdUyhIS2kvqu_1c-n5AkZAJYHIsp7hFm5_SrWFl8EvA8Pv9s4HHZu325qt0oyFukdiL_EBb4EKB4M8EfTVgWIAyWMYZ1yKu-gGrmLb6jSoa_-yMlB5aP709yAsS0z7DqHzKveD13MRHuM9FHj_lOXzrzRRPA/https%3A%2F%2Faka.ms%2FLearnAboutSenderIdentification CAUTION: External Sender
External Email - Use Caution
Dear FreeSurfer Community,
We are currently processing volumetric data from a longitudinal study (adolescent patients with anorexia nervosa vs. healthy controls, 2–3 time points) using FreeSurfer 8.0.0 and the standard longitudinal pipeline (recon-all -long <tpNid> <templateid> -all).
We performed Quality Control (QC) directly on the longitudinal outputs, rating the segmentation for each region individually. Based on this QC, we identified several areas requiring manual correction due to significant overestimation. (We are aware that manual corrections should be reserved for cases with substantial segmentation inaccuracies....)
According to the FreeSurfer documentation, manual edits should ideally be performed on the cross-sectional runs and the base template rather than the longitudinal files themselves. Upon inspecting our data, we observed a few conflicting points:
* Cross-sectional and base accuracy: In many cases, the segmentation in the cross-sectional outputs and the base images actually appears more accurate than in the longitudinal outputs, despite the perceived higher resolution of the latter.
* Lack of "Edit-Trigger": So, in several instances, the specific areas we flagged as problematic in the longitudinal output look fine in both the cross-sectional images and the base template. If there is nothing to "fix" in the earlier stages, re-running the longitudinal pipeline is unlikely to change the erroneous longitudinal output.
This leaves us with a methodological dilemma: Our current QC (identifying errors) is based on the longitudinal output. If we shift to editing cross-sectional images, we would essentially need to redo the QC for those runs, even though they already appear superior to the final product.
What is the recommended procedure in this case? Should we proceed with manual edits directly on the longitudinal outputs despite the documentation’s warnings?
Any insights or recommendations on the best practice for this scenario would be greatly appreciated.
Furthermore, we are questioning whether the application of a longitudinal pipeline is appropriate for our specific study population. Our sample includes adolescents undergoing normative neurodevelopmental processes, alongside patients with Anorexia Nervosa who exhibit significant brain volume reductions (approximately 5%) in the acute stage, followed by substantial recovery at follow-up. We are concerned that these pronounced morphological changes and the resulting high variance could potentially bias the creation of the within-subject base template, possibly affecting the accuracy of the longitudinal registration.
Best regards,
Lara
freesurfer@nmr.mgh.harvard.edu
-
Bahrani, Ahmed A. -
Keller, Lara