[Camino-users] pico

[Philip A Cook]

>> track -inputfile brainDT.Bdouble -schemefile direction.scheme -inputmodel bayesdirac_dt -seedfile mask -iterations 100   | procstreamlines -outputacm -header mask -outputroot bayes_


The Bayesian method takes raw data as input, it fits a restricted diffusion tensor model (with the _dt suffix) to estimate the likelihood function over the sphere. The correct call is:

  track -inputfile RAW_DATA.Bfloat -schemefile direction.scheme -inputmodel bayesdirac_dt -seedfile mask -iterations 100   | procstreamlines -outputacm -header mask -outputroot bayes_

Where RAW_DATA.Bfloat is your DWI data in voxel order, stored as floats (the output of image2voxel).



On Mar 18, 2013, at 4:16 AM, Silvia De Santis wrote:

>> Dear Philip,
>> 
>> many thanks for your suggestions. PDFs are ok, and so the seeds I think, because if I run track -inputmodel dt I get reasonable results. 
>> 
>> But as I try this:
>> 
>> track -inputfile brainDT.Bdouble -schemefile direction.scheme -inputmodel bayesdirac_dt -seedfile mask -iterations 100   | procstreamlines -outputacm -header mask -outputroot bayes_
>> 
>> I get the error:
>> 
>> SEVERE: Exception in thread "main" class data.DataSourceException: End of file reached without completing voxel.
>> 
>> I have the latest software version. Any suggestions?
>> 
>> Many thanks again,
>> 
>> Silvia
>> 
>> 
>> 
>> On Mar 4, 2013, at 6:37 PM, Philip A Cook wrote:
>> 
>>> Hi, 
>>> 
>>> A good first step is to look at the PDFs with pdview:
>>> 
>>>  pdview -inputmodel pico -inputfile brainPICoPDFs.Bdouble -datadims <x> <y> <z>
>>> 
>>> where x,y,z are the dimensions of your image. Check the orientations and concentration parameters are sensible. The image should resemble an FA image.
>>> 
>>> The next thing I would do is check the seeds are in the correct place. I usually check this by doing simple tracking first:
>>> 
>>>  track -inputfile  brainDT.Bdouble -inputmodel dt -seedfile mask | procstreamlines -outputacm -header mask -outputroot dt_streamline_
>>> 
>>> BTW, you can use the Bayesian tracking algorithm directly with track, skipping the PDF fitting process:
>>> 
>>>  track -inputfile data.Bfloat -schemefile imaging.scheme -inputmodel bayesdirac_dt -seedfile mask -iterations 100 | procstreamlines -outputacm -header mask -outputroot bayes_
>>> 
>>> 
>>> 
>>> On Mar 4, 2013, at 4:00 AM, Silvia De Santis wrote:
>>> 
>>>> Hello,
>>>> 
>>>> I am trying to use pico to estimate the probability of connectivity between a bunch of voxels and I get inconsistent results, so I was seeking for help to understand what I am doing wrong. 
>>>> 
>>>> cat brainDT.Bdouble | picopdfs -inputmodel dt -luts picoTable.dat > brainPICoPDFs.Bdouble
>>>> 
>>>> track -inputfile brainPICoPDFs.Bdouble -inputmodel pico -seedfile mask | procstreamlines -outputcp -outputroot results/ -seedfile mask  
>>>> 
>>>>> brainDT.Bdouble is obtained with dtfit and looks ok
>>>>> mask is an indexed mask of voxels (from 1 to n_voxels)
>>>>> picoTable is obtained with dtlutgen
>>>> 
>>>> I get .img files that basically contains white spots that are outside the brain, and since this is the first time I use camino I am really clueless.
>>>> 
>>>> Any kind of help would be greatly appreciated!
>>>> 
>>>> Silvia
>>>> 
>>>> 
>>>> *****************************************************
>>>> Silvia De Santis
>>>> CUBRIC
>>>> School of Psychology
>>>> Cardiff University
>>>> Tel: 029 2087 0354
>>>> Fax: 029 2087 0339
>>>> Web: http://psych.cf.ac.uk/contactsandpeople/researchfellows/desantis.php
>>>> *****************************************************
>>>> 
>>>> 
>>>> 
>>>> *****************************************************
>>>> Silvia De Santis
>>>> CUBRIC
>>>> School of Psychology
>>>> Cardiff University
>>>> Tel: 029 2087 0354
>>>> Fax: 029 2087 0339
>>>> Web: http://psych.cf.ac.uk/contactsandpeople/researchfellows/desantis.php
>>>> *****************************************************
>>>> 
>>>> _______________________________________________
>>>> Camino-users mailing list
>>>> Camino-users at www.nitrc.org
>>>> http://www.nitrc.org/mailman/listinfo/camino-users
>>> 
>> 
>> *****************************************************
>> Silvia De Santis
>> CUBRIC
>> School of Psychology
>> Cardiff University
>> Tel: 029 2087 0354
>> Fax: 029 2087 0339
>> Web: http://psych.cf.ac.uk/contactsandpeople/researchfellows/desantis.php
>> *****************************************************
>> 
> 
> *****************************************************
> Silvia De Santis
> CUBRIC
> School of Psychology
> Cardiff University
> Tel: 029 2087 0354
> Fax: 029 2087 0339
> Web: http://psych.cf.ac.uk/contactsandpeople/researchfellows/desantis.php
> *****************************************************
> 
> _______________________________________________
> Camino-users mailing list
> Camino-users at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/camino-users