Freesurfer August 2025

freesurfer@nmr.mgh.harvard.edu

31 participants
32 discussions

Large discrepancies between FreeSurfer v7.2.0 and v8.0.0 outputs
by Hana McMahon 26 Sep '25

26 Sep '25

External Email - Use Caution Hi, I’ve processed the same set of T1‐weighted scans with both FreeSurfer v7.2.0 and the newer v8.0.0. When I compare the regional stats (aseg + subfield tables), the absolute proportional differences are surprisingly large (anywhere from ~1 % up to nearly 200 %, depending on the structure). I first noticed the issue while running ANCOVAs: with the v8 outputs, estimated total intracranial volume (eTIV) is highly significant in every model, whereas the v7 results (using identical R code and covariates) show a very different pattern. I am wondering: * Are these version‐to‐version shifts expected, given the changes in v8? * Have others reported similar magnitude differences, particularly for eTIV and subcortical/hippocampal measures? * Are large shifts expected from run-to-run as well? * Any insight would be greatly appreciated, as I am concerned by the magnitude of the differences I am seeing between versions, which are significantly changing my results. Thanks for your time. Kindest, Hana Confidential communication - This email and any files transmitted with it are confidential and are intended solely for the addressee. If you are not the intended recipient, please be advised that you have received this email in error and that any use, dissemination, forwarding, printing, or copying of this email and any file attachments is strictly prohibited. If you have received this email in error, please notify me immediately by return email and destroy this email.

3 10

Question about using FreeSurfer license on another computer
by Andy Junghyun Kim 23 Sep '25

23 Sep '25

External Email - Use Caution Hi, I am currently using FreeSurfer on my local laptop and have a license.txt file. I will also need to run FreeSurfer on our lab server. Can I simply copy my existing license.txt file to the lab server, or do I need to request a new license for each computer? Kind regards, Andy Kim

2 2

21 Sep '25

External Email - Use Caution Hi Freesurfer community, I am currently trying to project 3D density maps in individual diffusion space onto the cortical surface (within individual space). I'm not sure which registration function to use to obtain the register.dat file required for mri_vol2surf . Does bbregister work for this situation? If so, assuming I create a registration from diffusion space (RASmm/ACPC) to fsnative, would the --mov file as the T1w in diffusion space – same affine as that of the diffusion data, and same resolution as diffusion data– (obtained from qsiprep) work with the target being the T1 in fsnative space? Or should I use another type of function such as tkregister2 or something like this? My current code is the following: template_file = op.join(bids_path, 'derivatives', 'qsiprep', participant, 'anat',f"{participant}_space-ACPC_desc-preproc_T1w.nii.gz") registration_file = op.join(out_dir, 'ACPC2fsnative_register.dat') # 1) Generate registration file .dat cmd = ["bbregister", "--s", participant, "--sd", fs_path, "--mov", template_file, "--reg", registration_file, "--init-fsl", "--t1" ] print("Running:", " ".join(cmd)) subprocess.run(cmd, check=True, env={**os.environ, "SUBJECTS_DIR": fs_path}) #2) Apply ACPC2fsNative registration and project density map to surface hemisphere = ['L', 'R'] roi = ['LeftMTxWMxLGN', 'RightMTxWMxLGN'] for h, hemi in enumerate(hemisphere): mask = roi[h] density_map = op.join(analysis_path, 'tdi_maps', 'dipy_tdi_maps', participant, f"{participant}_ses-concat_desc-{mask}_tdi_map.nii.gz") # density_img = nib.load(density_map) # print(density_img.affine) # Output file name out_file = os.path.join(out_dir, f"{participant}_hemi-{hemi} _space-fsnative_label-{mask}_tdi_on_surf.mgh") if hemi == 'L': hm = 'lh' else: hm = 'rh' # Build mri_vol2surf command cmd = [ "mri_vol2surf", "--mov", density_map, "--regheader", participant, "--hemi", hm, "--reg", registration_file, "--surf", "white", "--projfrac", "0", "--sd", fs_path, "--out", out_file ] # Run the command print("Running:", " ".join(cmd)) subprocess.run(cmd, check=True, env={**os.environ, "SUBJECTS_DIR": fs_path}) Thanks, Best, -Loïc

2 2

Zero Values for SegmentedTotalIntracranialVol
by Carly McIntyre-Wood 21 Sep '25

21 Sep '25

2 1

Trouble figuring out error
by Reyes, Nicholas 21 Sep '25

21 Sep '25

External Email - Use Caution Hello! I cannot seem to figure out the problem with this patient. I am newer to using freesurfer, so I am a little lost trying out what I've read in the documentation. Could you help give me some insight on figuring this out? Thank you in advance, Nicholas K. Reyes | B.S. in Neuroscience Neural Signals Processing and Modeling Lab The University of Alabama at Birmingham nkreyes(a)uab.edu<mailto:nkreyes@uab.edu> || 205-937-3421 Nicholas K. Reyes | UNP Major & BME Minor College of Arts and Sciences The University of Alabama at Birmingham nkreyes(a)uab.edu<mailto:nkreyes@uab.edu> || 205-937-3421

2 1

Issue with Skull Strip when Original Volume is Cut at the Nose
by 이준혁 21 Sep '25

21 Sep '25

External Email - Use Caution Environment - OS: Ubuntu 22.04.5 LTS - FreeSurfer: 7.4.1 I'm running FreeSurfer (Ubuntu 22.04 version 7.4.1), and during the Autorecon1 preprocessing, I noticed that the result depends on whether the original 3D volume includes the full face or the nose part is cut. When you run a truncated picture rather than a full face as input, the Skull strip results are completely different. In this case, do you have any official recommendations for normal preprocessing? How should I proceed I will attach the original picture and the pre-processing result picture.

2 1

Re: Surfa update
by Hoopes, Andrew 08 Sep '25

08 Sep '25

Thanks Chris, Good point. We can get on this. I’ll investigate putting in a quick fix. The requirement for numpy<2 stems from complications with building cython utilities. I think we should ditch these utilities and instead rely on prebuilt tools (eg scipy). It sacrifices some minor runtime, but will relieve major distribution headache. That project is a bit more involved but hopefully we can get to it soon. Andrew From: Rorden, Chris <RORDEN(a)mailbox.sc.edu> Date: Thursday, August 14, 2025 at 12:22 PM To: freesurfer(a)nmr.mgh.harvard.edu <freesurfer(a)nmr.mgh.harvard.edu>, Hoopes, Andrew <AHOOPES(a)mgh.harvard.edu> Subject: Surfa update External Email - Use Caution Hello Would it be possible to release a version of surfa that uses numpy 2? For examples, the original version of synthsr and the one included with FreeSurfer rely on Python versions that are end of life. I upgraded this to use up to Python 3.13 (latest version supported by TensorFlow), but the Surfa support demand for numpy 1 requires special environment so it does not play well with other tools (surfa pyproject.toml notes 'numpy<2.0’). I really do not want to maintain a separate fork, and if we could upgrade surfa the maintained FreeSurfer project could include versions of these scripts that support maintained versions of Python. For the time being, here is my fork: https://secure-web.cisco.com/1Tj_72h5VBPhEzQKtS8Wnrte2Od2rEUMf2iVThbaWZrFxe… Thanks for all your work, -chris

2 2

Version 7.2 & 7.4
by Taylor, Lisa 08 Sep '25

08 Sep '25

External Email - Use Caution Hi group, I was curious if using both versions of freesurfer 7.2 & 74 would be acceptable? We realized some of our data contained both due to some miscommunication. Would the recommendation be to use just one 7.X version, or are the two similar enough to use concurrently? Lisa

3 3

Cortical Thickness From Volume-Space ROIs - Initial Registration Slightly Misaligned
by Rowan Lawrence 27 Aug '25

27 Aug '25

External Email - Use Caution Hello FreeSurfers, We have a number of subjects on which we have run recon-all to completion. We have a number of ROIs for the parahippocampal gyrus (perirhinal, entorhinal, and parahippocampal cortices) defined in volume space on the original T1-weighted images which are registered well, and I would like to extract cortical thickness estimates from these ROIs. I have so far been following the page "Cortical Thickness of a Volume-defined ROI" (https://secure-web.cisco.com/1DVU7ZoSo7uvfZJSPChJydu5jwQ_zDu8idGFGJ4xbpLsX8…) I am using version 8.1.0. I registered the T1 image to the fsaverage image using: fslregister --s fsaverage --mov /mnt/d/.../T1_asl_CON_L01.nii --reg T1_asl_CON_L01_to_fsaverage.dat Which ran without issue. Viewing with the suggested command: tkregisterfv --mov /mnt/d/../T1_asl_CON_L01.nii --reg T1_asl_CON_L01_to_fsaverage.dat --surf orig The registration looks pretty good, although the target structures in the parahippocampal gyrus do not perfectly overlap. Continuing with mri_vol2surf: mri_vol2surf --mov /mnt/d/.../ROIs/T1_asl_CON_L01_L_MTLcROI.mgz --reg T1_asl_CON_L01_to_fsaverage.dat --projdist-max 0 1 0.1 --interp nearest --hemi lh --out lh.fsaverage.MTLc.mgh And then viewing using the new tksurferfv: tksurferfv fsaverage lh inflated -overlay lh.fsaverage.MTLc.mgh -fthresh 0.5 The ROI appears roughly in the right position, but does not fully overlap the correct gyrus and appears slightly misaligned. Using mri_surf2surf and then writing out the stats with segstats: mri_surf2surf --s T1_asl_CON_L01 --trgsubject fsaverage --hemi lh --sval lh.thickness --tval lh.thickness.fsaverage.mgh mri_segstats --seg lh.fsaverage.MTLc.mgh --in lh.thickness.fsaverage.mgh --sum segstats_MTLc_L.txt The mean thickness is 1mm or less for each of the structures, which does not seem to be correct as manually checking the thickness of parts of the ROIs using measurement tools like ITK-Snap indicates the more expected thickness of ~2.5 - 3.0mm. I assume this must be due to the slightly misaligned ROI when viewed on the fsaverage brain. I have checked other posts on the mailing list and found the suggestion to specify --dof 12 when running fslregister. However, this made the issue considerably worse with the ROI now located in the brainstem and in entirely the wrong orientation. I'm confused about why this has happened and how to fix either issue. Thank you, any help greatly appreciated! Best wishes, Rowan ~ GTA/PhD Candidate School of Health Sciences Booth St. East, 4.034 - M13 9SS

2 1

[FreeSurfer] Different results for T1 vs T2 preprocessing (recon-all 7.4.1, Ubuntu 22.04)
by 이준혁 25 Aug '25

25 Aug '25

External Email - Use Caution Hello FreeSurfer team, I am preprocessing the IXI dataset and noticed that my T2 results look incorrect, even when I run the same pipeline as for T1 with only the input changed to the T2 file. I would appreciate guidance on what I’m doing wrong or how I should process T2 properly. Environment - OS: Ubuntu 22.04.5 LTS - FreeSurfer: 7.4.1 - SUBJECTS_DIR: /mnt/e/freesurfer_subjects - FSF_OUTPUT_FORMAT: nii.gz (default unless noted) What I ran - For a single subject, I tried: input="/mnt/e/T2_all/12subject_12.nii.gz" sid="subject_12T2" outdir="/mnt/e/output" 1) recon-all stage 1 recon-all -parallel -i "$input" -autorecon1 -subjid "$sid" 2) Apply Talairach transform mri_convert "$SUBJECTS_DIR/$sid/mri/brainmask.mgz" --apply_transform "$SUBJECTS_DIR/$sid/mri/transforms/talairach.xfm" -o "$SUBJECTS_DIR/$sid/mri/brainmask_align.mgz" 3) Convert to NIfTI mri_convert "$SUBJECTS_DIR/$sid/mri/brainmask_align.mgz" "$outdir/${sid}_brainmask_align.nii.gz" Issue - The T2 result looks abnormal (while T1 looks OK). I assumed I could reuse the exact same steps for T2, just swapping the input file. - Additionally, when I follow the steps from this guide: https://secure-web.cisco.com/1E1tkmifDB72D9Pbo3i9gNVonTconZI613a_nQKLd5pOD7… my output NIfTI “info” (dimensions, sform/qform, etc.) is slightly different from the author’s preprocessed files (e.g., they have 160×192 while mine often become 256×256×256 after conformation). I’d like to understand why. Questions 1) Is it correct that FreeSurfer’s recon-all is primarily designed for T1, and that T2 should NOT be run through the same standalone pipeline? 2)Then, how can I preprocess T2 using the same code as T1? Is there another method within FreeSurfer to handle T2 properly? 3) “Could you explain why some of the ‘info’ fields differ between the author’s preprocessed files and my outputs, even though I followed the same code?” Thank you very much for your time and guidance! Best regards, [JHL]

1 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Freesurfer August 2025