Print

Print
Thankyou Tim and Saad,

I actually didn't save the directory of the failed exclusion mask attempt, but it was scripted, and I commented out the commands with exclusion so probtrackx.log should have looked exactly like the example...I shall try it again, starting tonight if I can, to confirm that I can replicate the results...perhaps I should try the 4.5 hour one again as well, to make sure it is still behaving and has not gone south with the others...It DOES seem that I have had bad times lately...obviously all of this may take a while, and I wanted to respond with what I had.

(Killed after ~ 20 hours...but was running on e3131)
probtrackx --network --mode=seedmask -x cst_l/masks.txt -l -c 0.2 -S 2000 --steplength=0.5 -P 5000 --avoid=cst_l_exclusion_diff.nii.gz --forcedir -f --opd -s Bed.bedpostX/merged -m Bed.bedpostX/nodif_brain_mask --dir=cst_l 
===========================
4.5 hours, NO exclusion mask:
[dti_erp/e3131/cst_l_f] dpat% cat probtrackx.log
probtrackx --network --mode=seedmask -x cst_l_f/masks.txt -l -c 0.2 -S 2000 --steplength=0.5 -P 5000 --forcedir -f --opd -s Bed.bedpostX/merged -m Bed.bedpostX/nodif_brain_mask --dir=cst_l_f
================================================================
And, for e3131, I did a direct comparison of probtrack with and without the -f flag:
With -f was 38 minutes longer...but both were in the 3-4 hour range

===============================================
Perhaps the machine is misbehaving in some way...as these results seem to make no sense in your experience : (
This is the quad core intel mac with the 64 bit patched universal binaries running on it and 4 gb of ram.

-Dianne


On 9/11/07, Tim Behrens <[log in to unmask]> wrote:
Hi - Saad and I are a bit baffled by this.  

Are you sure the difference is not with the --f flag. The --f flag will definitely have a major effect on run time, but the exclusion mask (--avoid) should not. This should only affect which streamlines are counted. 


 

Could you send us the probtrack.log files for the analysis that took 4.5hrs, and the analysis that took 20+ hours. 

T



On 11 Sep 2007, at 19:17, Dianne Patterson wrote:

Hmm...when I create an exclusion masks that completely covers the contralateral hemisphere, probtrackx slows to a crawl (and, in fact, I just killed it when it was still crawling along 20 hours later)....sad considering that when I don't use the exclusion mask, the same track took 4.5 hours..

-D

On 8/31/07, Steve Smith < [log in to unmask]> wrote:
Hi,

On 31 Aug 2007, at 23:41, Dianne Patterson wrote:

> Hi All,
> I'm trying to understand what factors add to the running time of
> probtrackx.
> Obviously the size of the multiple masks in the connectivity analysis
>
> 1) BUT, what about the size of the exclusion mask...will a large
> exclusion mask increase or decrease running time?

Decrease, I think but would depend on the topography of the mask.

> Additionally, if I want to keep fibers from crossing the hemisphere
> boundary, is there any difference between the behavior of the
> following 2 masks?
> A) A sheet or slice at midline that separates the hemispheres
> B) A mask that completely covers one of the hemispheres

I think they would have the same effect in this case.

Cheers.


>
> Thanks much,  and I hope everyone is enjoying the course.
>
> -Dianne
> --
> Dianne Patterson, Ph.D .
> [log in to unmask]
> ERP Lab
> University of Arizona
> 621-3256 (Office)


------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---




--
Dianne Patterson, Ph.D.
[log in to unmask]
ERP Lab
University of Arizona
621-3256 (Office)




--
Dianne Patterson, Ph.D.
[log in to unmask]
ERP Lab
University of Arizona
621-3256 (Office)