Freesurfer May 2025

freesurfer@nmr.mgh.harvard.edu

21 participants
24 discussions

Question on Samseg Basic and Samseg for MS lesions
by Stefano Cerri 15 May '25

15 May '25

External Email - Use Caution Hi Arianna, 1) Yes. Please note that the method was tuned and mostly tested on MS patients. 2) These two versions of SAMSEG model WMH/ WM lesions differently and cannot be directly compared. The basic SAMSEG is not designed for accurate WMH or WM lesion estimation. Note that you can use the lesion version of SAMSEG on scans without lesions (see Fig. 6 here <https://secure-web.cisco.com/1KM7GJ1i4uKDVvBOOHjfxLxbNpEbDV37ju32_ri0-iGtre…>). 3) Yes. Note that SWI data might not add valuable information to the lesion and whole-brain segmentation task. My suggestion is to visually inspect a couple of subjects with and without SWI data. Stefano Sent on behalf of Arianna Menardi: ---- Dear Experts, I'm trying to quantify the volume of some white matter lesions in an elderly group. I have tried using both Samseg cross-sectional, as well as Samseg for MS lesions. I am new to the approach so I have 3 main questions: 1. Can I use Samseg for MS lesions even if my patients present WM lesions but are not MS patients? 2. the estimated values of WM-hypointensities derived from the basic version of Samseg are almost half in size compared to the output 'Lesions" of Samseg specific for MS lesions. What is hence the difference between the two approaches that could explain such a big difference in the output? 3. Can I use Samseg combining T1w, FLAIR and SWI data? Or the algorithms are not suited for SWI images? Thank you for your help, Arianna The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Mass General Brigham Compliance HelpLine at https://secure-web.cisco.com/1Kb8pmdPdKdYdZdQofh1Zs9Y9vJBM9d8tTBbuN-azPwq2U… <https://secure-web.cisco.com/1tDmFFExvcidAJHTtPomz2TcTR-cmBcLvo3q_epfYYnELl…> . Please note that this e-mail is not secure (encrypted). If you do not wish to continue communication over unencrypted e-mail, please notify the sender of this message immediately. Continuing to send or respond to e-mail after receiving this message means you understand and accept this risk and wish to continue to communicate over unencrypted e-mail.

3 2

Striatum and cerebellum parcellation stats from Choi2012 and Buckner 2011
by Payam Sadeghi Shabestari 15 May '25

15 May '25

External Email - Use Caution Hi FS team, I'm trying to obtain anatomical statistics for the regions defined in the Choi 2012 and Buckner 2011 parcellations, but I couldn't figure out how to get stats on them. I already did: for mask_option in "${mask_options[@]}"; do mri_vol2vol --mov $SUBJECTS_DIR/$subject_id/mri/norm.mgz \ --s $subject_id \ --targ $SUBJECTS_DIR/MNI152/buckner_atlas/17_network_${mask_option}_mask.nii.gz \ --m3z $SUBJECTS_DIR/MNI152/mri/transforms/talairach.m3z \ --noDefM3zPath \ --o $SUBJECTS_DIR/$subject_id/mri/cerebellum_17_network_${mask_option}_mask.nii.gz \ --inv-morph \ --interp nearest Thank you and best regards, Payam

2 3

Hippocampal Subvolumes Error -- segmentHA_T1_long
by Toms, Robert 14 May '25

14 May '25

External Email - Use Caution Dear FreeSurfer Community, I'm longitudinally processing a significant amount of secondary scan data, to get hippocampal subvolumes for a study I'm working on. Of the hundreds of scans I've processed so far, 18 have failed on line 2136 of SegmentSubfieldsT1Longitudinal.m--a subprocess of the segmentHA_T1_long command-- all with the same final lines: ####### Fitting mesh to image data mask took, at this global iteration and time point, 20.7436 seconds Transforming points Transforming points Wrote image to file image_tp_1.mgz Computing shift introduced by kvlRead Constructing image-to-world transform from header information (asmr1.mgz) Constructing image-to-world transform from header information (/mnt/md0/ScratchPad_/ABCD/Package_1234721/image03/derivative/freesurfer/freesurfer8.0.0/NDARINVGCPTFXEV_long/tmp/hippoSF_T1_long.v22_right/imageDump.mgz) Wrote image to file asmr2.mgz This file does not contain MRI parameters This file does not contain MRI parameters Unable to perform assignment because the left and right sides have a different number of elements. Error in SegmentSubfieldsT1Longitudinal (line 2136) MATLAB:matrix:singleSubscriptNumelMismatch Command exited with non-zero status 249 ####### I've searched the list and no similar errors have been reported. Does anyone know what may be the problem? I'll attach a couple of the long-hippocampal-subfields.log's in case they can be of help. FreeSurfer version: 8.0.0 System: Ubuntu 22.04.5 LTS, 512GB RAM Thanks, Robert

3 4

Re: [Freesurfer] SegmentSubfieldsT1Longitudinal Error, Line 2136 -- Are disorganized volumes usable?
by Toms, Robert 14 May '25

14 May '25

External Email - Use Caution (please see below) ________________________________ From: Toms, Robert Sent: Monday, May 5, 2025 2:55 PM To: freesurfer(a)nmr.mgh.harvard.edu <freesurfer(a)nmr.mgh.harvard.edu> Subject: SegmentSubfieldsT1Longitudinal Error, Line 2136 -- Are disorganized volumes usable? Dear Experts, I'm having the same error as this 2023 post that y'all responded to: https://secure-web.cisco.com/15Vt3zMobf7Y4tHpCjLxvyPhoC2gjNR1_ZY3GyZjqm3L4s… I'm running FreeSurfer 8.0.0 on Ubuntu 22.04.5 LTS, and most of my subjects have run through the whole pipeline with no problems. You told the original 2023 poster that the 2 timepoint volumes are disorganized, but still usable—which was very helpful, and covers most of my error cases, however two of my errored-out subjects have 3 timepoints. I wanted to check with you in case 3-timepoint volumes wouldn't be usable for some reason? The following volumes are in base/tmp/hippoSF_T1_long.v22_right/ : volumesAmygdala_tp_1.txt volumesAmygdala_tp_2.txt volumesAmygdala_tp_3.txt volumesHippo_tp_1.txt volumesHippo_tp_2.txt volumesHippo_tp_3.txt I've attached the logs for both of them, and the final lines of both logs are identical: Applying LTAtransformInterp (resample_type 0) writing to discreteLabelsMergedBodyHeadNoMLorGCDGResampledT1_tp_2.mgz... Transforming points Transforming points Wrote image to file image_tp_3.mgz This file does not contain MRI parameters Unable to perform assignment because the left and right sides have a different number of elements. Error in SegmentSubfieldsT1Longitudinal (line 2136) Thank you! Robert

2 1

Re: [Freesurfer] Run NextBrain on a high-res 3D dataset: tweaks required for 200um on Edlow brain?
by Iglesias Gonzalez, Juan E. 14 May '25

14 May '25

Oooohhhh I see. Yeah you’ll have to run in CPU mode on a machine with quite a lot of RAM 😊 -- Juan Eugenio Iglesias http://www.jeiglesias.com From: Keenlyside, Andrew <andrew.keenlyside.23(a)ucl.ac.uk> Date: Wednesday, May 14, 2025 at 3:27 AM To: Iglesias Gonzalez, Juan E. <JIGLESIASGONZALEZ(a)mgh.harvard.edu>, Freesurfer support list <freesurfer(a)nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] Run NextBrain on a high-res 3D dataset: tweaks required for 200um on Edlow brain? External Email - Use Caution Dear Juan, This was run with the fast version of NextBrain using the CPU running on a 313um HiP-CT scan of the brain (29MB). The GPU version at this resolution failed when I was trying to run it on a fairly large workstation (A1000 GPU ~48GB VRAM) The issue is not quite the performance of the model at ~300um, we have issues with the running of a 150um scan which is killed during the fitting step. We were wondering which versions you used to get the 150um segmentation of the edlow et al. MRI brain? And if you have any advice which might help us here, Kind regards, Andrew Andrew Keenlyside PhD Student, UCL Mech. Eng. ________________________________ From: Iglesias Gonzalez, Juan E. <JIGLESIASGONZALEZ(a)mgh.harvard.edu> Sent: 14 May 2025 07:25 To: Freesurfer support list <freesurfer(a)nmr.mgh.harvard.edu> Cc: Keenlyside, Andrew <andrew.keenlyside.23(a)ucl.ac.uk> Subject: Re: [Freesurfer] Run NextBrain on a high-res 3D dataset: tweaks required for 200um on Edlow brain? ⚠ Caution: External sender Dear Matthieu, Which version did you use? The FireANTs version, I presume? Can you please share some screenshots of the incorrect registration? Cheers, /Eugenio -- Juan Eugenio Iglesias http://www.jeiglesias.com From: freesurfer-bounces(a)nmr.mgh.harvard.edu <freesurfer-bounces(a)nmr.mgh.harvard.edu> on behalf of Chourrout, Matthieu <m.chourrout(a)ucl.ac.uk> Date: Tuesday, May 13, 2025 at 6:17 AM To: freesurfer(a)nmr.mgh.harvard.edu <freesurfer(a)nmr.mgh.harvard.edu> Cc: Keenlyside, Andrew <andrew.keenlyside.23(a)ucl.ac.uk> Subject: [Freesurfer] Run NextBrain on a high-res 3D dataset: tweaks required for 200um on Edlow brain? External Email - Use Caution Hi Freesurfer Community, The NextBrain page showcases the results of the pipeline on the Edlow brain binned at 200um, but we weren't able to reproduce this on our data with a voxel size < 300 um (thus we binned the data). With my colleague Andrew, we've pinpointed the issue to the cartoon fitting step. On what system was the NextBrain run at this resolution? I assume it was run on CPU only? Did you have to tweak the cartoon or other key steps to make it run? Best regards, Dr Matthieu Chourrout [Image removed by sender. orcid logo 16px] <https://secure-web.cisco.com/1tiMIHDVZbi1_soxzRzk-v66TQ9fs4j0sOPE5nQKOghqmP…> Research Fellow on Correlative X-ray Imaging for the Large-scale Imaging of Neural Circuits UCL Mechanical Engineering<https://secure-web.cisco.com/1YgMFScJszy12Gw3eD79ViRNN_Bv5QtTSAxq4Qz7GI7Knh…> Materials, Structure and Manufacturing at Harwell (MSMaH)<https://secure-web.cisco.com/1BkqAWAaOy_nkoRqQ3vWb0R--mjPV-_kDCnT3nqiNmzBSV…> Rm 4.17 Malet Place Eng. Building, Torrington Place, London, WC1E 7JE Bloomsbury Campus, University College London The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Mass General Brigham Compliance HelpLine at https://www.massgeneralbrigham.org/complianceline<https://secure-web.cisco.com/1LR5lLvByLZ91PM8q5FyO8ubUZqDRw0XrSw1PUU05ESPDs…> . Please note that this e-mail is not secure (encrypted). If you do not wish to continue communication over unencrypted e-mail, please notify the sender of this message immediately. Continuing to send or respond to e-mail after receiving this message means you understand and accept this risk and wish to continue to communicate over unencrypted e-mail.

1 0

Run NextBrain on a high-res 3D dataset: tweaks required for 200um on Edlow brain?
by Chourrout, Matthieu 14 May '25

14 May '25

External Email - Use Caution Hi Freesurfer Community, The NextBrain page showcases the results of the pipeline on the Edlow brain binned at 200um, but we weren't able to reproduce this on our data with a voxel size < 300 um (thus we binned the data). With my colleague Andrew, we've pinpointed the issue to the cartoon fitting step. On what system was the NextBrain run at this resolution? I assume it was run on CPU only? Did you have to tweak the cartoon or other key steps to make it run? Best regards, Dr Matthieu Chourrout [orcid logo 16px] <https://secure-web.cisco.com/1tiMIHDVZbi1_soxzRzk-v66TQ9fs4j0sOPE5nQKOghqmP…> Research Fellow on Correlative X-ray Imaging for the Large-scale Imaging of Neural Circuits UCL Mechanical Engineering<https://secure-web.cisco.com/1YgMFScJszy12Gw3eD79ViRNN_Bv5QtTSAxq4Qz7GI7Knh…> Materials, Structure and Manufacturing at Harwell (MSMaH)<https://secure-web.cisco.com/1H0MGshh9I-h03eOLwUTo_X7Tm443MK0cYSkOlls2X_Meo….> Rm 4.17 Malet Place Eng. Building, Torrington Place, London, WC1E 7JE Bloomsbury Campus, University College London

2 1

freeview -ss orientation inconsistencies
by Ania Holubecki 13 May '25

13 May '25

External Email - Use Caution Hi, I have a script that utilizes the -ss flag on the freeview command to take axial, sagittal, and coronal screenshots iteratively. I have multiple images loaded into freeview at a time for each screenshot, and most of the loaded images are the same across all the screenshots, except for one spherical ROI that is only the same within a given sagittal/axial/coronal set. I have been encountering a problem where some of the screenshots are horizontally mirrored (left-hemisphere/right hemisphere flipped) in comparison to the rest of the images. This happens the most often for the coronal screenshot, but has also occurred for the axial screenshot. The spherical ROIs for which this mirroring occurs is not consistent - every time I run the script, it's a different set that's horizontally mirrored. Has anyone encountered this issue and has a way to guarantee the orientation of the screenshot? Any tips are greatly appreciated! Best, *Ania Holubecki*

2 1

Question on Samseg Basic and Samseg for MS lesions
by Sennott, Liam Ricker 12 May '25

12 May '25

Sent on behalf of Arianna Menardi: ---- Dear Experts, I'm trying to quantify the volume of some white matter lesions in an elderly group. I have tried using both Samseg cross-sectional, as well as Samseg for MS lesions. I am new to the approach so I have 3 main questions: 1. Can I use Samseg for MS lesions even if my patients present WM lesions but are not MS patients? 2. the estimated values of WM-hypointensities derived from the basic version of Samseg are almost half in size compared to the output 'Lesions" of Samseg specific for MS lesions. What is hence the difference between the two approaches that could explain such a big difference in the output? 3. Can I use Samseg combining T1w, FLAIR and SWI data? Or the algorithms are not suited for SWI images? Thank you for your help, Arianna

1 0

Re: [Freesurfer] mri_glmfit error - matrix ill-conditioned. no solution from archive.
by Douglas N. Greve 09 May '25

09 May '25

Not necessarily. This error can arise if any of the columns are linearly dependent. But try rescaling first On 5/9/2025 8:29 AM, Ane Gurtubay wrote: > > External Email - Use Caution > > Dear Douglas, > > whenever we get this error *ERROR: matrix is ill-conditioned or badly > scaled* as in > https://www.mail-archive.com/freesurfer%40nmr.mgh.harvard.edu/msg74674.html > <https://secure-web.cisco.com/1oJ3YaCat1otNMsv9b4ieBxS300R4fr4eMozJ0nxWe7N3b…>, > should we normalize the values in the fsgd file or the design matrix > (Xg.dat)? > > Thank you in advance, > Ane > > ---- > Ane Gurtubay Antolin, PhD > Ikerbasque Research Fellow > Basque Center on Cognition, Brain and Language (BCBL) > http://www.bcbl.eu/en/ane-gurtubay-antolin > <http://secure-web.cisco.com/10Mht_1YL1Yjet8TT5lrE72xv1nVXLPfSiYFb6jv4xo-YOw…> >

1 0

mri_cc segementation fault during recon-all using FreeSurfer 8.0.0
by 김근유(의과대학 정신과학교실) 06 May '25

06 May '25

External Email - Use Caution I'm attempting recon-all command with "-all -qcache" option with T1w nii.gz file. I got the following error in all of my subjects data. ------ freesurfer-macOS-darwin_arm64-8.0.0-20250205-0f2bd3a FreeSurferEnv.csh dev Darwin gimgeun-yuui-MacBookPro.local 23.1.0 Darwin Kernel Version 23.1.0: Mon Oct 9 21:27:24 PDT 2023; root:xnu-10002.41.9~6/RELEASE_ARM64_T6000 arm64 ~/Dropbox/KKY/Imaging/FS_subjects/05731016/mri ~/Dropbox/KKY/Imaging/FS_subjects/05731016/mri IsConformed yes Mon Apr 14 18:40:16 KST 2025 mri_cc -aseg aseg.auto_noCCseg.mgz -o aseg.auto.mgz -lta transforms/cc_up.lta 05731016 will read input aseg from aseg.auto_noCCseg.mgz writing aseg with cc labels to aseg.auto.mgz will write lta as transforms/cc_up.lta Segmentation fault ERROR: mri_cc -aseg aseg.auto_noCCseg.mgz -o aseg.auto.mgz -lta transforms/cc_up.lta 05731016 Darwin gimgeun-yuui-MacBookPro.local 23.1.0 Darwin Kernel Version 23.1.0: Mon Oct 9 21:27:24 PDT 2023; root:xnu-10002.41.9~6/RELEASE_ARM64_T6000 arm64 recon-all -s 05731016 exited with ERRORS at Mon Apr 14 18:40:17 KST 2025 To report a problem, see http://secure-web.cisco.com/1OIbiZUZzpMZLVXGKj-AwW4K8SnoS4EHxuX0Nm7vR74Z11F… ------ I've used Freesurfer version 7.4.1 without this error in my computer. "aseg.auto_noCCseg.mgz" was only 16 bytes. And when I inspected "synthseg.vol.csv" in "stats" folder, the volume of all segments was 0. [image: image.png] I've searched the list and no similar errors have been reported. Does anyone have any thoughts on how to trouble-shoot this one? Also, Ive attached the recon-all.log in case it's of any use. 1) FreeSurfer version: MacOS X darwin_arm64 (64b) PKG installer (Silicon Macs with M1...M4 processors) -- with the latest version of Xquartz 2) Platform: macOS Somoma 14.1 3) uname -a: Darwin gimgeun-yuui-MacBookPro.local 23.1.0 Darwin Kernel Version 23.1.0: Mon Oct 9 21:27:24 PDT 2023; root:xnu-10002.41.9~6/RELEASE_ARM64_T6000 arm64 4) recon-all.log: see attached Thank you for your cooperation. Best regards KY, Kim -- *Keun You, KIM (김근유), MD, PhD* Assistant Professor Department of Psychiatry (Geriatric Psychiatry) Severance Hospital Yonsei University College of Medicine Seoul, Republic of Korea

2 1

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Freesurfer May 2025