Freesurfer December 2024

freesurfer@nmr.mgh.harvard.edu

45 participants
62 discussions

Re: [Freesurfer] Question on using thalseg_1.1.h5
by Georgios Argyropoulos 23 May '25

23 May '25

External Email - Use Caution Many thanks and apologies, Henry - I figured it out... I stupidly replaced your improved segmentation script in both the scripts and the bin directory instead of only replacing the former. many thanks for your time, much appreciated g

2 5

Importing surface meshes from 3rd party cortical surface reconstruction tools for analysis in fsaverage template space
by James Hengenius 03 Apr '25

03 Apr '25

External Email - Use Caution Hi, I'm working with infant (0-30 month) MR data, and iBEAT ( https://secure-web.cisco.com/1T2WKIelhUu8n_wuNR_5gb37RydVg1xrWEVvlUOdcz2UFW…) is currently doing a better job than Infant FreeSurfer ( https://secure-web.cisco.com/1M3VIitCZu3azJ0yMsnC2UvSxSyhcRAkbfTg_v1IUkOJOb…) at cortical surface recon in this age range. However, I have custom surface ROI atlases on the fsaverage template and would very much like to leverage FreeSurfer's stats tools to map iBEAT subject surfaces to the template and extract stats from my custom ROIs. To do so, I need to - Bring my iBEAT VTK surfaces (~47K vertices per hemisphere) into FreeSurfer (mris_convert?) - Resample the meshes to FS surface resolution (~12.5K vertices per hemi) - Compute curvature stats on the iBEAT surfaces - Use mris_surf2surf (or equivalent) to map the resampled iBEAT surface to the fsaverage template using curvature - Compute/extract anatomical stats from the regions equivalent to the ROIs in my template space atlas. All of the above are presenting challenges. So far, I tried using mris_convert to bring the VTK into FS format, but when I tried using mris_anatomical_stats, I get errors. Note that I also ran InfantFS on the data, and that I copied the converted (but not downsampled) VTK pial surface from iBEAT to $SUBJECTS_DIR/$subjid/surf/lh.ibeat.pial > mris_anatomical_stats -log test.ibeat.log sub-0001 lh ibeat.pial outputting results to test.ibeat.log... reading volume /scratch/t/tpaus/jhengeni/iBEAT/surface_remapping/sub-0001/sub-0001/mri/wm.mgz... reading input surface /scratch/t/tpaus/jhengeni/iBEAT/surface_remapping/sub-0001/sub-0001/surf/lh.ibeat.pial... Using TH3 vertex volume calc Total face volume 58098.7 Total vertex volume 57390.1 (mask=0) reading input pial surface /scratch/t/tpaus/jhengeni/iBEAT/surface_remapping/sub-0001/sub-0001/surf/lh.pial... error: mrisReadTriangleFile opened /scratch/t/tpaus/jhengeni/iBEAT/surface_remapping/sub-0001/sub-0001/surf/lh.pial okay but surface doesn't match /scratch/t/tpaus/jhengeni/iBEAT/surface_remapping/sub-0001/sub-0001/surf/lh.ibeat.pial. nvertices:46854 != mris->nvertices:12240 || nfaces:93704 != mris->nfaces:24476 error: mrisReadTriangleFile failed. error: mris_anatomical_stats: could not read surface file /scratch/t/tpaus/jhengeni/iBEAT/surface_remapping/sub-0001/sub-0001/surf/lh.pial It appears this approach is failing due to a difference in meshes (the iBEAT mesh is at a different resolution, with more elements, than the lh.pial from InfantFS). Any tips for this and the other steps would be a big help! Thanks, Jim H. -- Jim Hengenius Research Associate Paus Research Group CHU Sainte-Justine Research Centre My Google Calendar <http://secure-web.cisco.com/1NyYdK1uBklqcDSsFMhBiedpJnPC-H6m83iYBtfLt-c-oKz…>

2 2

Assistance with mri_vol2surf: Missing Surface Values in Output
by Pandey, Shraddha 08 Jan '25

08 Jan '25

External Email - Use Caution Dear Free Surfer Experts, I am currently working on generating a surface from a volume using the mri_vol2surf tool. The process I am following is as follows: 1. I first create the registration file using bbregister with the following command: bash bbregister --s "${sub}xx${ses}" --mov "$input_mri" --reg "$output_reg_file" --o "$output_vol_file" --init-header --t1 2. Here, the goal is to generate the .dat registration file. 3. Next, I use this .dat file in mri_vol2surf to generate the surface with the following command: mri_vol2surf --mov "$input_volume" --reg "$output_reg_file" --hemi lh --projfrac 0.2 --o "$output_surface" The input MRI for the bbregister command is flash data, and the input volume for mri_vol2surf is a paramagnetic volume that has been registered with the Flash data. However, the surface output generated by mri_vol2surf contains missing surface values (as shown in the figure below. For instance, the grey regions in the surface estimates). I would greatly appreciate any insights into why this might be occurring and how to resolve the issue. Thank you for your time and assistance! Regards, Shraddha [cid:3033bc8c-0a28-47a6-aa01-2a43edc2a75b]

3 6

Issues with MRI and PET Registration to fsaverage Template
by 문준호 02 Jan '25

02 Jan '25

External Email - Use Caution Hello, I am currently working on registering MRI and PET data from the ADNI dataset, processed using our custom preprocessing pipeline, to the fsaverage template. My goal is to align both the MRI and PET data to the fsaverage template. To achieve this, I followed these steps: *Step 1:* I used mri_coreg to create an LTA file for registering the MRI data to the fsaverage template. Then, I used mri_vol2vol to perform the registration of the MRI data to the fsaverage template. Commands Used: mri_coreg --mov $MRI_PATH --targ $FREESURFER_HOME/subjects/fsaverage/mri/brain.mgz --reg ${OUTPUT_DIR}/mri${num}_reg.lta mri_vol2vol --mov $MRI_PATH --targ $FREESURFER_HOME/subjects/fsaverage/mri/brain.mgz --reg ${OUTPUT_DIR}/mri${num}_reg.lta --o ${OUTPUT_DIR}/mri${num}_to_atlas.nii.gz *Step 2:* I used the LTA file generated in Step 1 and applied mri_vol2vol to register the PET data to the MRI data. Commands Used: mri_vol2vol --mov $PET_PATH --targ ${OUTPUT_DIR}/mri${num} _to_atlas .nii.gz --reg ${OUTPUT_DIR}/mri${num}_reg.lta --o ${OUTPUT_DIR}/pet${num}_to_atlas.nii.gz However, when visualizing the results using Freeview, I noticed that the registration of both the MRI and PET data to the fsaverage template is not accurate. (The MRI and PET data were successfully co-registered.) I have attached some images of the results and the output from the mri_info command for your reference. # freeview results ## Example 1[image: 6f1a5a16-f77f-493c-8802-925ef33a7f98.png] ## Example 2 [image: image.png] ## Example 3 [image: 37a638c7-944d-424c-af84-4fa1860560dc.png] # mri_info results ## T1w MRI niiRead(): detected input as 64 bit double, reading in as 32 bit float Volume information for {my_mri_data} type: nii dimensions: 128 x 128 x 128 voxel sizes: 1.000000, 1.000000, 1.000000 type: FLOAT (3) fov: 128.000 dof: 1 xstart: -64.0, xend: 64.0 ystart: -64.0, yend: 64.0 zstart: -64.0, zend: 64.0 TR: 0.00 msec, TE: 0.00 msec, TI: 0.00 msec, flip angle: 0.00 degrees nframes: 1 PhEncDir: UNKNOWN FieldStrength: 0.000000 ras xform present xform info: x_r = -1.0000, y_r = -0.0000, z_r = 0.0000, c_r = -13.0000 : x_a = -0.0000, y_a = 1.0000, z_a = -0.0000, c_a = 16.5000 : x_s = 0.0000, y_s = 0.0000, z_s = 1.0000, c_s = 11.5000 Orientation : LAS Primary Slice Direction: axial voxel to ras transform: -1.0000 -0.0000 0.0000 51.0000 -0.0000 1.0000 -0.0000 -47.5000 0.0000 0.0000 1.0000 -52.5000 0.0000 0.0000 0.0000 1.0000 voxel-to-ras determinant -1 ras to voxel transform: -1.0000 -0.0000 -0.0000 51.0000 -0.0000 1.0000 -0.0000 47.5000 -0.0000 -0.0000 1.0000 52.5000 -0.0000 -0.0000 -0.0000 1.0000 ## brain.mgz Volume information for /usr/local/freesurfer/7.4.1/subjects/fsaverage/mri/brain.mgz type: MGH dimensions: 256 x 256 x 256 voxel sizes: 1.000000, 1.000000, 1.000000 type: UCHAR (0) fov: 256.000 dof: 0 xstart: -128.0, xend: 128.0 ystart: -128.0, yend: 128.0 zstart: -128.0, zend: 128.0 TR: 0.00 msec, TE: 0.00 msec, TI: 0.00 msec, flip angle: 0.00 degrees nframes: 1 PhEncDir: UNKNOWN FieldStrength: 0.000000 ras xform present xform info: x_r = -1.0000, y_r = 0.0000, z_r = 0.0000, c_r = 0.0000 : x_a = 0.0000, y_a = 0.0000, z_a = 1.0000, c_a = 0.0000 : x_s = 0.0000, y_s = -1.0000, z_s = 0.0000, c_s = 0.0000 talairach xfm : /usr/local/freesurfer/7.4.1/subjects/fsaverage/mri/transforms/talairach.xfm Orientation : LIA Primary Slice Direction: coronal voxel to ras transform: -1.0000 0.0000 0.0000 128.0000 0.0000 0.0000 1.0000 -128.0000 0.0000 -1.0000 0.0000 128.0000 0.0000 0.0000 0.0000 1.0000 voxel-to-ras determinant -1 ras to voxel transform: -1.0000 -0.0000 -0.0000 128.0000 -0.0000 -0.0000 -1.0000 128.0000 -0.0000 1.0000 -0.0000 128.0000 -0.0000 -0.0000 -0.0000 1.0000 Could you please help me identify any potential issues with my approach or the commands I used? I would greatly appreciate any guidance you can provide. Thanks a lot! Best regards, Junho Moon

2 2

(no subject)
by Tommaso Mariani 02 Jan '25

02 Jan '25

External Email - Use Caution Dear authors, I'm trying to reconstruct a model of a brain from an MRI, and after completion of recon-all (without any errors) the results are truly poor. By looking through your tutorials I am quite confident to say that the problem is either one of intensity normalisation or of white matter segmentation (possibly, both), as you can see in the images attached. My first attempt is to insert some control points, though I wonder if I might need instead to fix the segmentation with the brush tool. Could it be possible to receive some feedback about this issue? Another issue regards the freeview software, as whenever I move with the arrows from a slice to another the loaded surfaces suddenly disappear and won't reappear unless I delete and reopen them, and it makes it really hard to insert control points if I can't see the surfaces' outlines. Also, I was wondering if the computational cost of the autorecon2-cp is dependent significantly on the number of control points. The pictures shows the lh.white, rh.white files over the T1.mgz file and the wm.mgz and control.dat files over the brain mask.mgz file. Thanks in advance for your help, Tommaso Mariani

2 3

How Can I Track Coordinates from NIfTI Raw Files
by 권용현 01 Jan '25

01 Jan '25

External Email - Use Caution Hello, I am conducting research using ECoG. I want to accurately track the location of electrodes. I am inputting data from a NIfTI file (size: 512, 512, 360, 1). I already know the exact electrode location within the NIfTI file, e.g., (182, 274, 260). The x, y, z values of this coordinate correspond to the raw fMRI scan, with x ranging from 0-512, y from 0-512, and z from 0-360. I extracted the --vox2ras matrix from the 001.mgz file using mri_info. I attempted to project the coordinates from the NIfTI file onto the generated pial surface, but I found a significant discrepancy. What I need is not Talairach coordinates, but a precise transformation of the electrode coordinates in the NIfTI space to their exact location on the subject's pial surface. How can I achieve this? Additionally, I noticed that the origin (0,0,0) of the generated pial surface does not align with the anterior commissure (AC). I am curious about the coordinate system of the final mesh – is it simply the voxel-based x, y, z system, or is it in the RAS coordinate system? Thank you. Best regards, Yonghyeon Gwon -- Yongheyon Gwon Interdisciplinary Program in Neuroscience, Seoul National University, Human Brain Function Laboratory, Seoul National University Hospital Seoul, Korea E-mail: gngyong(a)hbf.re.kr , gyh0321(a)naver.com Phone: +82-10-9706-3814

2 1

FreeSurfer 8.0.0-beta: Stack smashing and normalization failures with 7T MP2RAGE data
by Michael Eyre 01 Jan '25

01 Jan '25

External Email - Use Caution Hello I'm processing 7T MP2RAGE data (0.65mm isotropic) with FreeSurfer 8.0.0-beta. I have processed 30 subjects successfully but am encountering two distinct issues for 3 subjects: 1. Stack smashing errors during topology correction for subjects with large defects (sub-06, sub-31): Copy*** stack smashing detected ***: terminated Command terminated by signal 6 This occurs in mris_fix_topology when processing defects with many vertices: sub-06: Left hemisphere, defect 0 with 22,492 vertices sub-31: Right hemisphere, defect 2 with 14,313 vertices sub-06 does not have any obvious brain abnormalities. sub-31 has some white matter abnormalities at the terminal zones of myelination (deep posterior white matter). 2. Normalization failure for another subject (sub-04): Copyerror: MRInormFindControlPoints: could not find enough control points error: MRInormFindControlPoints failed This subject has some very minor scattered white matter abnormalities (perivascular spaces). My recon-all command: export FS_ALLOW_DEEP=1 recon-all -all -subject ${SUBJECT} -sd ${WORK_PATH}/freesurfer8 -hires -i ${WORK_PATH}/${SUBJECT}_uni_masked.nii.gz -T2 ${WORK_PATH}/${SUBJECT}_t2w_registered.nii.gz -T2pial -expert ${EXPERT_FILE} -openmp 20 My EXPERT_FILE contains two lines: mris_inflate -n 100 mris_fix_topology -threads 1 Preprocessing: Data are processed with 3D Gibbs deringing, ANTS DenoiseImage, bias field correction and brain extraction prior to freesurfer. Approaches tried: I have tried running without the EXPERT_FILE, and also tried with additional recon-all flags -3T -gcut -norm1-n 4 -norm2-n 6, without success. Previous successful processing: The same subjects processed successfully with FreeSurfer 7.4.1. System info: Ubuntu 20.04, 64GB RAM, 20 CPUs. Subjects are running in serial not in parallel, and no other processes with significant memory or CPU use are running on the system at the same time. Any suggestions would be greatly appreciated. I have attached the recon-all.log files. Let me know if you need any additional information. Best wishes Mike Eyre Dr Michael Eyre Locum Consultant Paediatric Neurologist, Evelina London Children's Hospital Visiting Research Fellow, School of Biomedical Engineering & Imaging Sciences, King's College London St Thomas' Hospital London SE1 7EH

2 1

Reliability of Optic Chiasm Volume Measurements in FreeSurfer
by Knut J Bjuland 01 Jan '25

01 Jan '25

External Email - Use Caution Dear FreeSurfer experts, I am examining the optic chiasm segmentation in FreeSurfer. How reliable is the volume measurement of the optic chiasm when using FreeSurfer? Best regards Knut Jørgen Bjuland

2 1

Bug Report for Beta Version of FreeSurfer (8.0.0-beta)
by Zhang, Chaokai 31 Dec '24

31 Dec '24

External Email - Use Caution Hello FreeSurfer Developers, Happy New Year! I am currently using the latest beta version of FreeSurfer (freesurfer-linux-ubuntu22_x86_64-8.0.0-beta) and have been impressed by the significant acceleration in the recon-all process. However, I encountered an issue while performing a simple labelization on a T1-weighted sample using mri_watershed. When running the command: mri_watershed -LABEL T1.mgz T1_label.mgz I noticed that the output file (T1_label.mgz) is shifted and does not align with the original T1-weighted sample. To confirm, I ran the same command using the stable version (7.4.1), and the output aligned correctly with the T1-weighted sample. [cid:image001.png@01DB5BA1.780AAF30] This is just a bug report and does not require any assistance. I wanted to bring this to your attention. Thank you for your hard work and dedication to improving FreeSurfer. details of my system: 1) FreeSurfer version: freesurfer-linux-ubuntu22_x86_64-8.0.0-beta 2) Platform: Ubuntu 24.04.1 LTS 3) uname -a: Linux Kai 5.15.167.4-microsoft-standard-WSL2 #1 SMP Tue Nov 5 00:21:55 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux Best regards, Chaokai

1 0

Issues with MRI and PET Registration to fsaverage Template
by 문준호 30 Dec '24

30 Dec '24

External Email - Use Caution Hello, I am currently working on registering MRI and PET data from the ADNI dataset, processed using our custom preprocessing pipeline, to the fsaverage template. My goal is to align both the MRI and PET data to the fsaverage template. To achieve this, I followed these steps: Step 1: I used mri_coreg to create an LTA file for registering the MRI data to the fsaverage template. Then, I used mri_vol2vol to perform the registration of the MRI data to the fsaverage template. Commands Used: mri_coreg --mov $MRI_PATH --targ $FREESURFER_HOME/subjects/fsaverage/mri/brain.mgz --reg ${OUTPUT_DIR}/mri${num}_reg.lta mri_vol2vol --mov $MRI_PATH --targ $FREESURFER_HOME/subjects/fsaverage/mri/brain.mgz --reg ${OUTPUT_DIR}/mri${num}_reg.lta --o ${OUTPUT_DIR}/mri${num}_to_atlas.nii.gz Step 2: I used the LTA file generated in Step 1 and applied mri_vol2vol to register the PET data to the MRI data. Commands Used: mri_vol2vol --mov $PET_PATH --targ ${OUTPUT_DIR}/mri${num}_to_atlas .nii.gz --reg ${OUTPUT_DIR}/mri${num}_reg.lta --o ${OUTPUT_DIR}/pet${num}_to_atlas.nii.gz However, when visualizing the results using Freeview, I noticed that the registration of both the MRI and PET data to the fsaverage template is not accurate. (The MRI and PET data were successfully co-registered.) I have attached some images of the results and the output from the mri_info command for your reference. # mri_info results ## T1w MRI niiRead(): detected input as 64 bit double, reading in as 32 bit float Volume information for {my_mri_data} type: nii dimensions: 128 x 128 x 128 voxel sizes: 1.000000, 1.000000, 1.000000 type: FLOAT (3) fov: 128.000 dof: 1 xstart: -64.0, xend: 64.0 ystart: -64.0, yend: 64.0 zstart: -64.0, zend: 64.0 TR: 0.00 msec, TE: 0.00 msec, TI: 0.00 msec, flip angle: 0.00 degrees nframes: 1 PhEncDir: UNKNOWN FieldStrength: 0.000000 ras xform present xform info: x_r = -1.0000, y_r = -0.0000, z_r = 0.0000, c_r = -13.0000 : x_a = -0.0000, y_a = 1.0000, z_a = -0.0000, c_a = 16.5000 : x_s = 0.0000, y_s = 0.0000, z_s = 1.0000, c_s = 11.5000 Orientation : LAS Primary Slice Direction: axial voxel to ras transform: -1.0000 -0.0000 0.0000 51.0000 -0.0000 1.0000 -0.0000 -47.5000 0.0000 0.0000 1.0000 -52.5000 0.0000 0.0000 0.0000 1.0000 voxel-to-ras determinant -1 ras to voxel transform: -1.0000 -0.0000 -0.0000 51.0000 -0.0000 1.0000 -0.0000 47.5000 -0.0000 -0.0000 1.0000 52.5000 -0.0000 -0.0000 -0.0000 1.0000 ## brain.mgz Volume information for /usr/local/freesurfer/7.4.1/subjects/fsaverage/mri/brain.mgz type: MGH dimensions: 256 x 256 x 256 voxel sizes: 1.000000, 1.000000, 1.000000 type: UCHAR (0) fov: 256.000 dof: 0 xstart: -128.0, xend: 128.0 ystart: -128.0, yend: 128.0 zstart: -128.0, zend: 128.0 TR: 0.00 msec, TE: 0.00 msec, TI: 0.00 msec, flip angle: 0.00 degrees nframes: 1 PhEncDir: UNKNOWN FieldStrength: 0.000000 ras xform present xform info: x_r = -1.0000, y_r = 0.0000, z_r = 0.0000, c_r = 0.0000 : x_a = 0.0000, y_a = 0.0000, z_a = 1.0000, c_a = 0.0000 : x_s = 0.0000, y_s = -1.0000, z_s = 0.0000, c_s = 0.0000 talairach xfm : /usr/local/freesurfer/7.4.1/subjects/fsaverage/mri/transforms/talairach.xfm Orientation : LIA Primary Slice Direction: coronal voxel to ras transform: -1.0000 0.0000 0.0000 128.0000 0.0000 0.0000 1.0000 -128.0000 0.0000 -1.0000 0.0000 128.0000 0.0000 0.0000 0.0000 1.0000 voxel-to-ras determinant -1 ras to voxel transform: -1.0000 -0.0000 -0.0000 128.0000 -0.0000 -0.0000 -1.0000 128.0000 -0.0000 1.0000 -0.0000 128.0000 -0.0000 -0.0000 -0.0000 1.0000 Could you please help me identify any potential issues with my approach or the commands I used? I would greatly appreciate any guidance you can provide. Thanks a lot! Best regards, Junho Moon

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 December 2024