[Camino-users] PICo parameters

[Philip A Cook]

Hi,

For single tensor tractography, I recommend using the Bayesian tractography algorithm. It does not require calibration, and will let you use -bgmask to limit the tracking to certain areas. To do this, you would pass the raw data to the track command, for example:

  track -inputfile data.Bfloat -inputmodel bayesdirac -header DTI.nii.gz -seedfile normROI83.nii.gz -iterations 1000 -curvethresh 90 -bgmask brain_mask.nii.gz -outputfile tracks83.Bfloat

If you really need to use calibration-based PICo, pdview will tell you if your LUT is sensible. It should somewhat resemble an RGB-encoded FA map, with higher concentration parameters in white matter. The usual cause of bad results is not having a scheme file in SI units (see the man page for dtlutgen for how to get around this) or not having a good estimate of SNR (see man page for estimatesnr). 

The -bgmask option only usually applies when handling DWI data, however you can pass a binary mask to track with -anisfile mask.nii.gz -anisthresh 1.0


Phil

On Jun 18, 2012, at 11:17 AM, Curci Marina Carmela wrote:

> Good morning,
> 
> I continue to have problems with PICo tractography on DTI data set.
> I've tried also to change parameters (lrange, step, model pdf) on the LUT but I don't see a lot of improvements on the final output. I will attach you 2 pictures: one is the probability map in which the seedfile ROI is the brain stem (probmap_default_acm_sc.nii.gz) and one is the probability map in which the seedfile ROI is the precentral region (probmap51_default_acm_sc.nii.gz).
> The model for the pdf is a bingham function, lrange is 1-10 and the step 0.02 as the default conditions. SNR is 30.
> Could you give your opinion about them please?
> 
> I've read that I can use the command pdview to view the outputs of picopdfs and I've tried to see them:
> 
> pdview -inputfile pdf_default.Bdouble -inputmodel pico -e1 -pdf bingham -datadims 96 96 34
> 
> For each pixel I see 3 different values. Are they the 3 components of the major eigenvalue? Then in brackets I have another number that is the sum of the absolute value of the concentration parameters in the voxel.
> How can I check from this last value if my lut is good or not? Will it be more probable that a track will cross voxels where this value is higher? Which parameter most influence this value?
> 
> And finally the last question: I run the track command using white matter mask as a background mask, but I see from the output that the background mask is not considered during the tracking. Why does this happen?
> 
> Waiting for your answers,
> I thank you,
> Marina
> 
> ________________________________________
> Da: Curci Marina Carmela
> Inviato: sabato 9 giugno 2012 9.01
> A: camino-users at www.nitrc.org
> Oggetto: PICo parameters
> 
> Good morning, I need a suggestion!
> 
> I have to do PICo probabilistic tractography on a DTI reconstructed data-set. Seeds are placed in a ROI involving the brain stem. I run the command track and then procstreamlines as reported below:
> 
> track -inputfile picopdf.Bdouble -inputmodel pico -header DTI.nii.gz -seedfile normROI83.nii.gz -iterations 5000 - curvethresh 90 -bgmask brain_mask.nii.gz -outputfile tracks83.Bfloat
> 
> procstreamlines -inputfile tracks83.Bfloat -outputacm -mintractlength 8 -seedfile normROI83.nii.gz -iterations 5000 -resamplestepsize 0.25 -outputroot probmapROI83
> 
> The resulting probability map is not so good in my opinion, because I expect that fibers start from the brain stem and they grow up in a fan-shaped way. Instead in my case the fibers go up only in the two main directions.
> 
> What do you think about it? Should I change some parameters running these commands? Or should I change something in the look-up table? I choose a Bingham pdf function, as suggested on the man page of dtlutgen.
> 
> Waiting for your feedback,
> I thank you for your helpfulness,
> Marina!
> _______________________________________________
> Camino-users mailing list
> Camino-users at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/camino-users