Freesurfer

freesurfer@nmr.mgh.harvard.edu

16 participants
30065 discussions

processing steps
by claudia wheeler-kingshott 25 Sep '06

25 Sep '06

Hi, I have just started to look at freesurfer and decided to run each step to try to understand what recon-all does. Well, after a bit of teething to put the data in the right orientation, we managed to start, but we have a few questions :-) First of all, our SPGR data is 1.2x1.2x1.2mm^3 and it is acquired coronally. I used mri_convert to get the data resampled to 1x1x1mm^3. 1) When I run the mri_nu_correct command to generate the nu file I get a lot of "Not yet installed" errors. Do I have to generate this file at all? It seems redundant. 2) When I get to the generation of the T1 file and then the brainmask, this last file is not compatible with the original data or the nu data because of a different voxel size. This causes me the first problem with the mri_em_register command, which fails if I use the --mask option with my brainmask image and the nu data, giving me a segmentation fault. If I use the T1 data, then the command runs fine! Is there a reason why we can't use the T1 file instead of the nu or orig file? 3) We also have a problem with the commands that use the $FREESURFER_HOME/average/RB40_talairach_2005-12-30.gca file because it is not in the directory. Can we use the RB_all_2006-02-15.gca file instead? What is the difference between these two files? 4) Has anybody got a feeling on how freesurfer works on brains with MS lesions? 5) Is there any way to find descriptions and info on commands that do not have an official page? Sorry for the basic questions and I hope to hear from you soon. Claudia

3 2

Getting PET statistics using the subcortical segmentation
by jorge luis 25 Sep '06

25 Sep '06

Hello all I have PET images corregistered in spm with their corresponding MRI images obtained by converting brain.mgz to analyze. I have created in freesurfer the file register.dat for each one of them, so i can use tksurfer and tkmedit. Now I would like to compute for each PET image the average value of the pixels intensities in several regions resulting from the subcortical segmentation (for example hippocampus) and others manually painted. Also I would like to produce an average region from all the individuals for each one of these selected regions (the average subject does not have a subcortical segmentation and it could be used as a mask in spm analysis). Any tip is welcome In advance Thank you ______________________________________________ LLama Gratis a cualquier PC del Mundo. Llamadas a fijos y móviles desde 1 céntimo por minuto. http://es.voice.yahoo.com

2 1

anonymizing dicom files
by Najmeh Khalili M. 25 Sep '06

25 Sep '06

sorry about nonstop posting! I'm trying to remove the patient names from dicom files. I was advised to use dcm_modify_elements but it fails on DCM_Openfile. Is there a straight forward way of doing this? Thanks Naj P.S. Doug, I am still looking for the graddir tags :)

8 10

Performance
by Raksha Urs 25 Sep '06

25 Sep '06

Hello We are running freesurfer on a Power Mac G4 with a 1.25 GHz processor and 1 GB Ram. The subcortical segmentation procedure recon-all subcortseg -s <subjid> ran for 49 hours. The rest of autorecon2 (recon-all -autorecon2-cp -s <subjid>) has been running for 6 days and is still working on the left hemisphere. Is this normal? Do you have any recommendations to speed up this process? Thank you Raksha Urs Research Associate Wien Center for Alzheimer's Disease and Memory Disorders Mount Sinai Medical Center, Miami Beach FL

2 1

surface and volume measurements
by pgreenberg＠ucsd.edu 22 Sep '06

22 Sep '06

Hi, 1) I'm interested in the relationship between voxels labeled within the aseg.mgz file and those bounded by the white matter and pial surfaces. In regions such as the hippocampus, the pial surface (red) may cut through the middle of the gray matter designated as hippocampus in the aseg volume overlay. It therefore looks like some of the hippocampus fits between the white/pial boundary and some lies outside the pial boundary. If I wanted to know hippocampal volume, would the value in the aseg stats file be correct, or would regions of the hippocampus outside the pial boundary not be included? 2) Ignoring the hippocampus and looking at the neocortex only, there again seems to be a difference between the volume of tissue found between the white/pial boundaries and that designated as either R or L hemisphere cerebral cortex in the aseg volume overlay. Specifically, in some of our data sets there is unlabeled gray matter to be found between the white/pial surfaces. If a voxel doesn't have an "aseg" label will it be counted toward the total volume in the aseg stats file? If that same voxel lies within the white/pial surfaces will it be counted as gray matter in a different stats file? Which output will provide the most accurate determination of neocortical gray matter volume? Thanks very much, Paul

2 1

strange segmented image data
by Frederick Klauschen 21 Sep '06

21 Sep '06

Hi, when loading the segmentated data into Matlab I get an image containing four smaller subimages (see attachment) instead of one (fullsize) that corresponds to the original input dataset. Does anybody know what went wrong? Frederick __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 1

regdat2xfm problem
by Stefano Pellegrini 20 Sep '06

20 Sep '06

Hi everybody, I have a problem with regdat2xfm. I have a reg.dat file that I made using tkregister2 to register a couple of volumes (targ .mgz and mov .img) from one of my subjects. Now, if I try to use the reg.dat file with tkregister2 and the two original files, everything is fine: I ask tkregister2 to read the reg file and the registration is what I made before. Then I convert reg.dat in reg.xfm, and I want to use the .xfm file with mincresample (after having converted the mov .img file in a .mnc file with mri_convert), I get a completely wrong transformation. also, I was wondering if there is anyway to write the output transformed volume directly inside tkregister2. these are the commands I am using data: targ.mgz mov.img (the file I want to move onto targ.mgz) commands: tkregister2 --targ targ.mgz --mov mov.img --regheaders --reg reg.dat regdat2xfm WM.img targ.mgz reg.dat reg.xfm mri_convert targ.mgz targ.mnc mri_convert WM.img WM.mnc mincresample -transformation reg.xfm -like targ.mnc WM.mnc reg_WM.mnc thanks, -stefano

3 4

Coordinate system clarification please
by Graham Wideman 20 Sep '06

20 Sep '06

Folks: I am (still!) trying to get traction on some aspects of the current FS coordinate system(s). In the recon-all pipeline, am I right in assuming that from orig.mgz all the way through to aseg.mgz and then on to brain.mgz, wm.mgz and so on, that there's NO resampling to a new coordinate system (eg: the specific brain's data doesn't get resampled to rotate or talairach it)? Ie: if you overlay, for example, wm.mgz's data onto orig.mgz, it'll align properly? (Ie: overlay the raw voxel data, unmolested by any clever viewer transforms...) I *think* I'm right on this, but it's hard to be sure, because for example, mri_ca_normalize's docs say that it takes a gca transform file as input. But that *might* be so that mri_ca_normalize can resample some other standard file to match the specific brainmask.mgz file. <scratching of head> Graham

4 4

Quadrangle surface formats
by Graham Wideman 18 Sep '06

18 Sep '06

Probably for Nick or Bruce? Is it the case that the quadrangle surface formats are obsolete, and only the triangle surface is currently used? Thanks, Graham

2 1

FW: field sign and colour lables
by kristofer kinsey 16 Sep '06

16 Sep '06

I have a question regarding colour labelling for the retinotopic maps generated by Freesurfer. I'm running Freesurfer-Linux-rh9-stable-pub-v3.0.2 and have successfully generated the polar, eccen, and field sign maps (no errors). The images seem to differentiate the occipital regions, except that the colour labelling appears to be incorrect. In the eccen image, for example, the order of the colours start out as red, green, blue and using the Configure Overlay Display Inverse it's green, blue, red. According to my notes one should see fovea=red, parafoveal=blue and periphery=green. Also, with the field sign image, the area that should be V1 shows up as blue instead of yellow (can't seem to invert the colours). I've since gone back to double check that my procedure is correct and re-checked my registration and can't seem to find my mistake. I was hoping to get some help to fix this problem. In case it helps, I've attached a few images of the maps that were generated. Cheers kk ______________________________ Kristofer Kinsey, PhD Neurosciences Research Institute School of Life and Health Sciences Aston University Birmingham B4 7ET Web Site: www.physiol.ox.ac.uk/~kk3/

2 1

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