Hi Tracula team I am using tracula to study the corticospinal tract in a patient population. I am wondering if there would be a way to have the different DTI metrics from the whole tract at different depth/portion of the tracts. Something like the information given by the pathstats.byvoxel.txt file, but for the whole width of the tract? If not, do you see a way to sample the value of the diffusion metrics on a certain tract starting from the cortex (like with projfrac or projdist with the WM surface)? Thanks very much for your help Celine
Dear FS-experts,
I've been using optseq for many years without problems, however, I run into a couple now. My code is the following:
optseq2 --ntp 255 --tr 2.0 --tprescan 4.0 --psdwin 0.0 16.0 1 --ev Gain2 5.0 8.0 --ev Gain4 5.0 8.0 --ev Gain6 5.0 8.0 --ev Gain8 5.0 8.0 --ev Loss2 5.0 8.0 --ev Loss4 5.0 8.0 --ev Loss6 5.0 8.0 --ev Loss8 5.0 8.0 --ev Catch 5.0 8.0 --tnullmin 2.0 --tnullmax 6.0 --tsearch 24 --focb 100 --ar1 .37 --evc 1 1 1 1 -1 -1 -1 -1 0 --evc -1.5 -.5 .5 1.5 -1.5 -.5 .5 1.5 0 --evc -1.5 -.5 .5 1.5 1.5 .5 -.5 -1.5 0 --cost eff --nkeep 3 --o output_run1
If I run this on a computational cluster (CentOS 5) with more than enough RAM, I get the following message after on screen summary: outstem = output_run1 *** glibc detected *** optseq2: malloc(): memory corruption: 0x0000000003f3e4e0 *** *** glibc detected *** optseq2: malloc(): memory corruption: 0x0000000003f3e4e0 ***
And nothing happens after that.
If I run the same code on my iMac Lion 10.7.5, it seems to be working, but I get multiple "warnings": /usr/pubsw/packages/vxl/1.13.0/src/core/vnl/algo/vnl_qr.txx: vnl_qr<T>::solve() : matrix is rank-deficient by 112
The number at the end changes each time. In addition, the process consumes my RAM (8GB) like it's nothing.
The strange thing is, I do get usable output.
Any ideas what might be causing this?
Many thanks!
Cédric
------------------------------------------------------------ P.C.M.P. Koolschijn (Cédric), PhD Dutch Autism & ADHD Research Center Brain and Cognition Amsterdam, The Netherlands E P.C.M.P.Koolschijn@uva.nlmailto:P.C.M.P.Koolschijn@uva.nl W http://www.dutcharc.nl
I think you are at the limit of what is possible. You have TR=2 and Ntp=255 for a total scan time of 510sec. You have 9 event types, each 5stim+2null=7sec long, each repeated 8 times. This calls for 504 for stimulation and so only 6 sec total left for jitter. doug
On 05/13/2013 05:28 AM, Koolschijn, Cédric wrote:
Dear FS-experts,
I've been using optseq for many years without problems, however, I run into a couple now. My code is the following:
optseq2 --ntp 255 --tr 2.0 --tprescan 4.0 --psdwin 0.0 16.0 1 --ev Gain2 5.0 8.0 --ev Gain4 5.0 8.0 --ev Gain6 5.0 8.0 --ev Gain8 5.0 8.0 --ev Loss2 5.0 8.0 --ev Loss4 5.0 8.0 --ev Loss6 5.0 8.0 --ev Loss8 5.0 8.0 --ev Catch 5.0 8.0 --tnullmin 2.0 --tnullmax 6.0 --tsearch 24 --focb 100 --ar1 .37 --evc 1 1 1 1 -1 -1 -1 -1 0 --evc -1.5 -.5 .5 1.5 -1.5 -.5 .5 1.5 0 --evc -1.5 -.5 .5 1.5 1.5 .5 -.5 -1.5 0 --cost eff --nkeep 3 --o output_run1
If I run this on a computational cluster (CentOS 5) with more than enough RAM, I get the following message after on screen summary: outstem = output_run1 *** glibc detected *** optseq2: malloc(): memory corruption: 0x0000000003f3e4e0 *** *** glibc detected *** optseq2: malloc(): memory corruption: 0x0000000003f3e4e0 ***
And nothing happens after that.
If I run the same code on my iMac Lion 10.7.5, it seems to be working, but I get multiple "warnings": /usr/pubsw/packages/vxl/1.13.0/src/core/vnl/algo/vnl_qr.txx: vnl_qr<T>::solve() : matrix is rank-deficient by 112
The number at the end changes each time. In addition, the process consumes my RAM (8GB) like it's nothing.
The strange thing is, I do get usable output.
Any ideas what might be causing this?
Many thanks!
Cédric
P.C.M.P. Koolschijn (Cédric), PhD Dutch Autism & ADHD Research Center Brain and Cognition Amsterdam, The Netherlands /E /P.C.M.P.Koolschijn@uva.nl mailto:P.C.M.P.Koolschijn@uva.nl /W /http://www.dutcharc.nl
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Celine - Since 5.2, there are 2 types of measures in the pathstats.byvoxel.txt file: The value at every point along the trajectory sampled at roughly the center of the bundle, or the value at every point along the trajectory averaged over the width of the bundle. Sorry, there are no other measures available at this point. If I understand correctly, you would like to see something like a profile of the measure of interest at every point along the trajectory?
a.y
On Mon, 6 May 2013, celine@nmr.mgh.harvard.edu wrote:
Hi Tracula team I am using tracula to study the corticospinal tract in a patient population. I am wondering if there would be a way to have the different DTI metrics from the whole tract at different depth/portion of the tracts. Something like the information given by the pathstats.byvoxel.txt file, but for the whole width of the tract? If not, do you see a way to sample the value of the diffusion metrics on a certain tract starting from the cortex (like with projfrac or projdist with the WM surface)? Thanks very much for your help Celine
Hi Anastasia I am very sorry I didn't see this email so it may be a doubloon with my question I just post. Indeed I am interested in a profile of the DTI metrics along the tract. So if I take the average value of each section it can give me a kind of profile, and so I was just wondering if taking a weighted metrics would be possible in order to remove values that are a bit far from the highest probability of the tract. Thanks a lot! Celine
Hi Celine - Since 5.2, there are 2 types of measures in the pathstats.byvoxel.txt file: The value at every point along the trajectory sampled at roughly the center of the bundle, or the value at every point along the trajectory averaged over the width of the bundle. Sorry, there are no other measures available at this point. If I understand correctly, you would like to see something like a profile of the measure of interest at every point along the trajectory?
a.y
On Mon, 6 May 2013, celine@nmr.mgh.harvard.edu wrote:
Hi Tracula team I am using tracula to study the corticospinal tract in a patient population. I am wondering if there would be a way to have the different DTI metrics from the whole tract at different depth/portion of the tracts. Something like the information given by the pathstats.byvoxel.txt file, but for the whole width of the tract? If not, do you see a way to sample the value of the diffusion metrics on a certain tract starting from the cortex (like with projfrac or projdist with the WM surface)? Thanks very much for your help Celine
Hi Celine - The way this average is computed for pathstats.byvoxel.txt is by averaging the values for each one of the sample paths that are drawn by the MCM algorithm. This essentially "weighted" sinced voxels that more sample paths go through will be counted more than voxels that fewer sample paths go through.
a.y
On Fri, 17 May 2013, celine@nmr.mgh.harvard.edu wrote:
Hi Anastasia I am very sorry I didn't see this email so it may be a doubloon with my question I just post. Indeed I am interested in a profile of the DTI metrics along the tract. So if I take the average value of each section it can give me a kind of profile, and so I was just wondering if taking a weighted metrics would be possible in order to remove values that are a bit far from the highest probability of the tract. Thanks a lot! Celine
Hi Celine - Since 5.2, there are 2 types of measures in the pathstats.byvoxel.txt file: The value at every point along the trajectory sampled at roughly the center of the bundle, or the value at every point along the trajectory averaged over the width of the bundle. Sorry, there are no other measures available at this point. If I understand correctly, you would like to see something like a profile of the measure of interest at every point along the trajectory?
a.y
On Mon, 6 May 2013, celine@nmr.mgh.harvard.edu wrote:
Hi Tracula team I am using tracula to study the corticospinal tract in a patient population. I am wondering if there would be a way to have the different DTI metrics from the whole tract at different depth/portion of the tracts. Something like the information given by the pathstats.byvoxel.txt file, but for the whole width of the tract? If not, do you see a way to sample the value of the diffusion metrics on a certain tract starting from the cortex (like with projfrac or projdist with the WM surface)? Thanks very much for your help Celine
freesurfer@nmr.mgh.harvard.edu
-
Anastasia Yendiki -
celine＠nmr.mgh.harvard.edu -
Douglas N Greve -
Koolschijn, Cédric