Freesurfer

freesurfer@nmr.mgh.harvard.edu

6 participants
30055 discussions

Subject: query about mri_vol2surf
by Pandey, Shraddha 17 Dec '24

17 Dec '24

External Email - Use Caution Hello FreeSurfer Experts, I am using mri_vol2surf to map a volume onto a surface, with the registration .dat file generated via bbregister. I would like to understand how the input volume is being sampled to generate the surface and how to visualize the sampled points. mri_vol2surf --src "$input_volume" --out "$output_surface" --srcreg "$output_reg_file" --hemi lh --projfrac 0.2 Specifically, how can I display the input_volume along with the sampled points at a projfrac of 0.2 on Freeview? Thank you in advance for your help! Shraddha

2 3

amygdala nuclei from segmentHA_T1 registration issue
by Stephanie Williams 17 Dec '24

17 Dec '24

External Email - Use Caution The amygdala nuclei labels from segmentHA_T1.sh seem to behave differently in standard freesurfer registration commands than other labels from aparc+aseg.mgz: I can't get the amygdala nuclei to register to my functional space correctly despite no issues with aparc+aseg labels. Are additional steps needed to correct header or orientation information when using amygdala nuclei labels? I'm extracting labels from one of the amygdala FSVoxelSpace files. Below is an example of the command I use to extract labels + register to functional space that isn't working. These commands work great with labels from aparc+aseg.mgz but end up in completely incorrect positions for any of the segmentHA-T1 amygdala nuclei. There is no error message, the issue is just that the output masks are in the wrong location. mri_extract_label rh.hippoAmygLabels.HBT.FSvoxelSpace.mgz 7001 labels/nuclei1_rh.nii mri_vol2vol --mov myfunc --targ labels/nuclei1_rh.nii --reg func2anat.dat --inv --o nuclei1_funcspace.nii The label file from the extract label command looks correct when overlaid on the anatomical. The issue seems to arise after vol2vol command -- the output mask is in the wrong area (eg amygdala nuclei are in the cerebellumafter vol2vol). The same vol2vol command works fine with the whole amygdala label from aparc+aseg, + I have checked the registration .dat file from bbregister is correct. Any advice would be appreciated! I've checked the header information using fslhd and the orientations from masks that worked (aparc+aseg) and didn't work (rh.hippoAmygLabels) appear identical. *I've abbreviated the name of the input to mri_extract_label command for simplification, the full name is: rh.hippoAmygLabels-T1.v22.HBT.FSvoxelSpace.mgz

1 0

Issues with hippocampal segmentation code (segmentHA_T2) and missing ConcatenateSubregionsResults in 7.3.1
by Chumin, Evgeny Jenya 17 Dec '24

17 Dec '24

External Email - Use Caution Hello, I am attempting to run hippocampal segmentation with freesurfer 7.3.1, using the segmentHA_T2.sh script with a high-resolution hippocampus T2 sequence as additional scan input. The code runs fine, but oddly it places the outputs into the mri subfolder rather than into stats (output file names shown below). mri/ rh.amygNucVolumes-T2HighResHippo.v22.txt rh.hippoSfVolumes-T2HighResHippo.v22.txt rh.hippoAmygLabels-T2HighResHippo.v22.CA.FSvoxelSpace.mgz rh.hippoAmygLabels-T2HighResHippo.v22.CA.mgz rh.hippoAmygLabels-T2HighResHippo.v22.FS60.FSvoxelSpace.mgz rh.hippoAmygLabels-T2HighResHippo.v22.FS60.mgz rh.hippoAmygLabels-T2HighResHippo.v22.FSvoxelSpace.mgz rh.hippoAmygLabels-T2HighResHippo.v22.HBT.FSvoxelSpace.mgz rh.hippoAmygLabels-T2HighResHippo.v22.HBT.mgz rh.hippoAmygLabels-T2HighResHippo.v22.mgz The stats folder also contains output .stats files for hippocampus and amygdala, but they only have one sentence in them: " # Amygdala nuclei volumes as created by segmentHA_T2.sh " These are the names of the files in the stats directory: stats/ amygdalar-nuclei.lh.T2.v22.T2HighResHippo.stats amygdalar-nuclei.rh.T2.v22.T2HighResHippo.stats hipposubfields.lh.T2.v22.T2HighResHippo.stats hipposubfields.rh.T2.v22.T2HighResHippo.stats It seems to me that the output files are initialized in the stats directory, but the values themselves get written to files in the mri directory. When I go to compile the data across subjects, the ConcatenateSubregionsResults script is missing from the installation (I have confirmed this both for 7.3.1 and 7.4.1 installations). I found the following thread in archive with a similar issue of this file missing: https://secure-web.cisco.com/1D6pNu3LKBDnQ_g29D1J9DrkaQ4h3xi6d3NHKg1r8YUImC… In this thread Antoine shared a direct download for the ConcatenateSubregionsResults, then Jackson shares a batched version of the script to resolve an issue with it looking for files in the /mri directory (which is where those files are written in my case anyways). Either by using the unbatched script or using the batched script and manually moving the files from the mri to the stats, I am able to run the script but run into another issue. The output file contains in the first row the values for the first subject and then for all subsequent subject rows it has region labels. I looked at each subject's output and it looks like this: Hippocampal_tail 547.477135 subiculum-body 198.862009 CA1-body 108.611407 subiculum-head 162.383267 hippocampal-fissure 132.385149 presubiculum-head 100.360234 CA1-head 477.119333 presubiculum-body 85.674081 parasubiculum 58.297345 molecular_layer_HP-head 180.316933 molecular_layer_HP-body 231.129446 GC-ML-DG-head 132.306221 CA3-body 66.096743 GC-ML-DG-body 115.076547 CA4-head 108.334300 CA4-body 105.624384 fimbria 86.764321 CA3-head 89.481195 HATA 52.204101 Whole_hippocampal_body 997.838938 Whole_hippocampal_head 1360.802930 Whole_hippocampus 2906.119003 My guess is that the labels are being incorrectly pulled from the second column and the values from the first column. Doe anyone have any insight on how to resolve this? I could always pull what I need manually, but if this is an issue with the code it would be nice to see it resolved. Many thanks! Jenya C. Indiana University School of Medicine

1 0

needed: tools to work with synthmorph deformations in ANTS and FSL pipelines.
by Josh 17 Dec '24

17 Dec '24

External Email - Use Caution mri_synthmorph is simply an amazing SOTA tool that I want to incorporate into all my pipelines. However, I am unsure how to convert the combined affine+nonlinear deformation image into a format that can be worked with by ants or FSL. Optimally, synthmorph + lta_convert + mri_convert would provide tools to: 1. separate affine and nonlinear components of the registration (useful in certain applications) 2. convert freesurfer transforms to current standard formats (FSL and ITK ants) -Josh

2 3

Single voxel grey-white matter boundary binary mask
by Tamir Eisenstein 16 Dec '24

16 Dec '24

External Email - Use Caution Dear FreeSurfer experts, Is it possible using the FS tools to create a grey-white matter boundary binary mask in volume space (e.g., a nifti file) that will be a single voxel thick along the entire grey-white matter boundary? Many thanks, Tamir Dr Tamir Eisenstein, Ph.D. Postdoctoral Researcher FMRIB, Wellcome Centre for Integrative Neuroimaging Nuffield Department of Clinical Neurosciences John Radcliffe Hospital, Oxford OX3 9DU University of Oxford tamir.eisenstein(a)ndcn.ox.ac.uk<mailto:julien.fars@ndcn.ox.ac.uk> eisentamir(a)gmail.com<mailto:julien.fars@gmail.com> tel: +44 7938790763

3 2

FreeSurfer Labels Question
by Barsotti, Ercole 13 Dec '24

13 Dec '24

External Email - Use Caution Hi all, Sorry if this is a stupid question. In the FreeSurferColorLUT.txt, there is a label for "Unknown" brain material. What would most commonly be labelled as Unknown in an image? Thanks, John

2 2

synthmorph for functional registration?
by Josh 13 Dec '24

13 Dec '24

External Email - Use Caution bbregister is known to be really could at registering functionals to T1w image. I am curious as to whether synthmorph performs close to bbregister or not when registering epi to the structural.

3 2

Spiky Brains
by Bos, M.G.N. (Marieke) 13 Dec '24

13 Dec '24

External Email - Use Caution Hello FreeSurfer Developers, We are completing quality control in FreeView of two longitudinal datasets (participants age: 9-25) and have noticed that in one dataset the brains look much more 'spiky' than in the other dataset. More than half of the brains have some 'spikes', < 10% has severe spikes; the most obvious ones correspond with both underestimations and overestimations in frontal regions of the brain (e.g., frontal pole, rostral middle frontal and superior frontal). The datasets are collected on two different scanners: Dataset with the spiky brains: * Structural MRI data were acquired using a Philips Achieva 3.0 Tesla MRI scanner with a standard whole-head coil. The following parameters were used: FOV = 224 mm x 178.5 mm x 168 mm; TR = 9.8 ms; TE = 4.6 ms; flip angle = 8 degrees; voxel size = 0.875 mm x 0.875 mm x 0.875 mm. The duration of the anatomical scan was 297.8 s. To reduce head motion, foam inserts were used and a film was projected on a screen. Dataset without spiky brains: * Structural MRI data were acquired using a Philips Achieva 3.0 TX MRI scanner with a standard whole-head coil. The following parameters were used: FOV = 250 mm × 196 mm × 170 mm; TR = 7.9 ms; TE = 3.5 ms, flip angle = 8 degrees, 228 × 177 × 155 slices; voxel size = 1.1 × 1.1 × 1.1 mm. The duration of the anatomical scan was 251 s. To reduce head motion, foam inserts were used. We are not too sure what to do about these. Manual skull stripping, adjusting the watershed threshold and using control points are not perfect solutions for various reasons. And we have noticed that the spikiness is reduced in some brains when the scans are processed in version 6.0 compared to version 7.3 - but otherwise version 7.3 seems like the better fit. We couldn't find anything in the archives, but we wonder - have you come across this before? Do you have another idea for us to try? Or is it possible that data from these regions would be (mostly) unimpacted by the appearance of small, infrequent spikes? If we want to use global brain measures, do you think Freesurfer gives a reliable estimation or would you advise us to exclude them from the analyses due to bad data quality. We have attached one picture of severe spikes so you know what we mean. We have started to rate the spikes from 1 (small/infrequent) to 4 (large/frequent). Thank you in advance for any ideas you can share with us. Best wishes, Manon Marieke

2 2

recon-all on poor MRI quality
by Francesco Lomi 13 Dec '24

13 Dec '24

External Email - Use Caution Dear all, I am writing you because I would like to know why recon-all on this subject exited with an error (I have uploaded the log file so you can see). Is it due to the poor quality of the T1 image? How to overcome this issue? I've added the FLAIR sequence to improve the surface reconstruction but the recon-all process did not end correctly. If you had any suggestions, they would be highly appreciated Best, Francesco

2 2

Freesurfer v8.0.0 error
by 潘晨盛 12 Dec '24

12 Dec '24

External Email - Use Caution Hello FreeSurfer Developers, I installed freesurfer 8.0.0 successfully but failed to run "recon-all -i *** -s *** -all -localGI -qcache" for any subjects. I tried this version 8.0.0beta on both macos latest version and ubuntu 22.04, the error was the same as follows: "Adding options -fix-ento-wm -transfer-base-bfs -fix-vsinus -fix-mca-dura -fix-ga -fix-acj -synthstrip -synthseg -synthmorph" "ERROR: cannot use ML routines" The command line did not even start anything, thus no log file was generated. I've searched the list and no similar errors have been reported. Does anyone have any thoughts on how to trouble-shoot this one?

3 2

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Freesurfer