Freesurfer April 2023

freesurfer@nmr.mgh.harvard.edu

80 participants
105 discussions

covariates in 2nd level anaysis
by Maria Czarnecka 01 May '23

01 May '23

External Email - Use Caution Dear Sisters and Brothers in neuroscience I am analyzing functional data in FSfast in Freesurfer. I define my conditions in an osgm file, then I create contrasts. On the second level, I would like to add two covariates (scanning facility and handedness). I don't need contrasts, I just need the model to account for the variance. Can I define both osgm and fsgd files? How do I construct the fsgd? I will be grateful for your suggestions. Best, Maria Czarnecka

2 1

Pass code for FS_7_2_0_Ubuntu_18_04_06.ova.7z
by Nathan Santana Schunk 30 Apr '23

30 Apr '23

External Email - Use Caution Hi, I would like to request the pass code needed to extract the FS_7_2_0_Ubuntu_18_04_06.ova.7z archive. Thanks in advance.

1 0

Fwd: questions about mri_normalize
by Fischl, Bruce R.,PHD 28 Apr '23

28 Apr '23

External Email - Use Caution Hi John 1. We empirically set it at 2, but that was a long time ago, but the optimal number is dependent on a lot of things, mostly image CNR. If the gray/white contrast is good enough you can run it for more, but you have to make sure that the normalization doesn’t start creeping into the gray matter. It is essentially a coarse segmentation that tries to find the bulk of the wm, then normalize it to 110. In regions of low gray/white CNR it would sometimes start normalizing the deeper layers to 110 at which point the thickness and surface placements will of course be wrong. 2. Yes, if the wm is not around 110 then something went wrong. You can also just put a single control point in the body of the wm, but I don’t think I’ve seen it not center most of the wm on 110, so maybe something else is going on? Cheers Bruce External Email - Use Caution Dear experts, I have been using the mri_normalize in FSv7.3.2 to standardize the intensity of FLAIR images , and I've observed that increasing the sigma parameter up to 25 is effective in improving the quality of the normalized images for noisy datasets. mri_normalize -sigma 25 -g 1 FLAIR_skullstripped.nii FLAIR_skullstripped_norm.mgz However, I have questions about the n argument, which is set to 2 by default. For noisy images, white matter intensity after normalization is around 140-150. When using n=10, I noticed that the white matter intensity decreased from 140-150 to around 110-120, resulting in better quality normalized images. 1) I'm wondering if there is a cut-off value for the number of iterations, or if a higher number of iterations always yields better results. 2) Also, if the intensity of the white matter after normalization is not around 110, but instead around 140 or 150, does that indicate that the normalization was not effective and that we need to increase the number of iterations? Thank you for your assistance. John

1 0

substantia nigra
by Lund, Anina 28 Apr '23

28 Apr '23

External Email - Use Caution Hi, Is it currently possible to extract metrics (volume, thickness, or area) of the substantia nigra with freesurfer?

2 1

parallel recon-all using T1,T2 images
by Toshiharu Kamishikiryo 28 Apr '23

28 Apr '23

External Email - Use Caution Hello Freesurfer experts I am trying to recon-all using T2 images as well to improve pial surfaces. Does anyone know how to use GNU Parallel in bash to run the following commands in parallel? recon-all -subject subjectname -i /path/to/T1w.nii -T2 /path/to/T2w.nii -T2pial -all -qcache Thanks in advance. Regards Kamishikiryo Research Student

3 3

Any large stash of labelled brain MRIs for CNN training?
by Keith E Moore 28 Apr '23

28 Apr '23

External Email - Use Caution Hi, First samseg and mri_reg are amazing tools. Thank you! I am running them on an internal dataset of images to create labelled MRIs. This is taking 45 min to an hour per grouping (T1, T1ce, T2 and FLAIR) to align and segment. It made me wonder if there is a cache of already labelled MRI images. I'm particularly looking for caudate and thalamus segmentation. Hoping I can leverage what others have done while I work on labelling my dataset. Keith

2 1

Freesurfer mailing list submissions
by julia emily 28 Apr '23

28 Apr '23

External Email - Use Caution

1 0

Automatic Calculation of White matter ADC Values using SynthSeg
by bar gen 28 Apr '23

28 Apr '23

External Email - Use Caution Dear Freesurfer experts, I hope this email finds you well. I am writing to seek your advice on a specific issue regarding a pediatric brain tumor and control group dataset that I am working with. The dataset consists of diffusion-weighted images (DWIs) with T2, b0, and b1000 trace images. My goal is to automatically calculate the Apparent Diffusion Coefficient (ADC) values in the contralateral white matter of both the childhood tumor and control groups. I have considered using b0 images segmented by SynthSeg to obtain the white matter images and then perform the measurement. However, I am not sure if this is the correct approach. When I use SynthSeg, the initial segmentation output has a resolution of 1x1x1 mm, while the ADC images have a resolution of 1.5x1.5x5 mm. I am unsure how to overcome this discrepancy in resolution. I would appreciate any guidance on whether using SynthSeg for segmenting the b0 images is the correct approach, as well as any suggestions for addressing the resolution discrepancy between the segmentation and ADC images. Additionally, if there are alternative methods or tools for automatically calculating ADC values in the contralateral white matter, I would be grateful for your recommendations. Kind regards. Barış

2 1

Fw: repeated measures anova for 4 treatments by 2 timepoints
by Resnick, Patricia 28 Apr '23

28 Apr '23

External Email - Use Caution Tricia Stefancin Lab manager Z-lab Department of Psychology The Ohio State University ________________________________ From: Resnick, Patricia Sent: Friday, April 21, 2023 12:29 PM To: freesurfer(a)nmr.mgh.harvard.edu Subject: repeated measures anova for 4 treatments by 2 timepoints Hello, I have a total of 8 subjects with 4 treatments (A/stim, A/sham, B/stim, B/sham) each with 2 timepoints (pre, post). I subtracted the pre-post images and I am trying to compare the differences between A/stim and A/sham, B/stim and B/sham, and A/stim and B/sham. I have run the mri_glmfit command below and get the following error: matrix is ill-conditioned or badly scaled, condno = 1.33391e+07 mri_glmfit --glmdir /fs/project/PAS1305/IMPACT/glm_analysis/motor/Motor-Fix/lh/ --y /fs/project/PAS1305/IMPACT/glm_analysis/motor/Motor-Fix/lh/sigLH.nii.gz --fsgd /fs/project/PAS1305/IMPACT/glm_analysis/motor/repeated_measures.fsgd doss --C /fs/project/PAS1305/IMPACT/glm_analysis/motor/TMS_stim-sham_repeated.mtx --C /fs/project/PAS1305/IMPACT/glm_analysis/motor/PBM_stim-sham_repeated.mtx --C /fs/project/PAS1305/IMPACT/glm_analysis/motor/PBM-TMS_stim_repeated.mtx --surf fsaverage lh --fwhm 0 --sim mc-full 500 2 corrp --sim-sign abs Would someone be able to tell me why this is not going through? Best, Tricia Stefancin Lab manager Z-lab Department of Psychology The Ohio State University

2 1

creating average labels
by Carlos Alcocer 28 Apr '23

28 Apr '23

External Email - Use Caution Good afternoon, I would like to verify that the subcallosal cingulate cortex (SCC) segmentation of my subjects are correct. I want to do this by creating an average label of the left SCC and an average label of the right SCC which I can then map onto each subject. If I save these labels from each subject as a volume (.mgz) and then run mri_concat on the left SCC labels (.mgz) and then on the right SCC labels (.mgz), would this accomplish my desired goal? Tks, Carlos

3 3

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Freesurfer April 2023