Hi Marc

On Mon, 19 Dec 2005, Marc Seal wrote:
> 
> a) what threshold would you recommend running this command at?  I'm  
> not sure what values are contained in these output images but the  
> fdt_paths.img have values in the range from 0 to 27605.

Hi - in the path images, these values represent the number of samples 
from each seed point that reached each voxel. If you have multiple seed 
points, the values just add. Hence, if you had a single seed point, and 
5000 samples, then  value/5000 is the prbability of that voxel being on 
the dominant connected path. However, these values obviously depend on a 
number of arbritrary things (such as voxel size - larger voxels have a 
higher probability of including the dominant path), and are therefore very 
hard to interpret generally - (contrast this to the connectivity-based 
segmentation option, where the targets are anatomically meaningful, and 
therefore the probabilities are also). It is best to look at an image or 
two, and try different thresholds. You can then ask yourself if each 
defines a sensible ROI. We usually find that small thresholds (~50/5000) 
reduce noise substantially.


> 
> b) what does the "-bin" operator do in this context and is it looking  
> for something specific in the output directory?  I have been writing  
> the ProdTrack output to a new directory since it tends to write over  
> the contents of existing directories.  Do I need anything else but  
> the fdt_path.img and relevant dti_FA.img?
> 

No - the bin operator just binarises the image before the muiltiplication 
to ensure everything inside the mask is 1.

Just the fdt_path and dti_FA should be fine (obviously you need the .hdr 
files also!)


> c) I'm assuming that the process of extracting FA values is the same  
> for output of all the ProbTack operations - is this correct? That is,  
> once I crack this I should be able to extract FA values from the  
> output of all the ProbTrack operations (Seed Voxel, Seed Masks, Two  
> masks - symmetric ...).
> 

Yep - this will work well for all these - you may need to think a little 
bit if you want to use it for seed-classification mode, but it generalises 
without too much trouble..

T



> 
> Many thanks & Merry Christmas,
> 
> cheers, Marc
> 
> 
> 
> ---------------------------------------------------------
> 
> Melbourne Neuropsychiatry Centre
> c/o  National Neuroscience Facility
> University of Melbourne
> 161 Barry Street
> Carlton South    VIC 3053
> Australia
> 
> ph:    (+61 3)  8344 1861
> fax:   (+61 3)  9348 0469
> 
> www.psychiatry.unimelb.edu.au/mh/mnc/
> www.brainmap.co.uk/right.htm
> 
> ----------------------------------------------------------
> 
> 
> 
> 

-- 
-------------------------------------------------------------------------------
Tim Behrens
Centre for Functional MRI of the Brain
The John Radcliffe Hospital
Headley Way Oxford OX3 9DU
Oxford University
Work 01865 222782
Mobile 07980 884537
-------------------------------------------------------------------------------