I have another, related question.
We are interested in the IFO, so we took a mask in the frontal lobe and another in the occipital lobe.
We tracked from frontal to occipital, and excluded low prob values based on a certain threshold of waytotal (say 1% for standard space, non-distance corrected data), and binarized the tract above this threshold
We then repeated the same process for the occipital-to-frontal tracking, with the same threshold of waytotal.
We still found that the remaining tracts had excess fibers, so we added an AND condition when we combined the two (fronto-occipital and occipito-frontal) representations of the tract.
Is this a correct procedure?
Stefano Marenco, MD
NIMH/CBDB
10 Center Drive, Bldg 10 room 3C103
Bethesda MD 20892
Tel 301 435-8964
Fax 301 480-7795
Email: [log in to unmask]
-----Original Message-----
From: Saad Jbabdi [mailto:[log in to unmask]]
Sent: Saturday, January 12, 2013 6:33 AM
To: [log in to unmask]
Subject: Re: [FSL] probtrackx thresholding
Hi
As has been discussed before on this list, it is very difficult to define generic rules for thresholding tractography results. The best approach will necessarily be tailored to the specific connections that you are looking at.
I would advise you to follow these steps (the first one is the tricky one):
1) Define inclusion/exclusion masks that allow you to accurately track the connections that you are interested in
2) Choose an arbitrary (but low) threshold to remove noise - this threshold should be relative to the value stored in the waytotal file
3) Average the results across subjects to define group tracts
Cheers
Saad
On 11 Jan 2013, at 21:43, Hellyer, Peter wrote:
> Hi All,
>
> Apologies in advance if this is obvious or obscure, just having a slight crisis of confidence on some probtrack work.
>
> I've been running probtrackx on a small group of subjects (12) to make a point to point connectivity matrix between a few regions. i.e. A->B,C,D, B->A,C,D, C->A,B,D etc resulting in a 4x4 connectivity matrix and 12 path distributions.
>
> I want to use the mean fdt_paths for each 'connection' across the group to sample FA from an independent group of subjects. For some of the resulting mean tracts, the maximum value in the mean distribution is well below 1, whereas in other tracts, the maximum value is several thousand.
>
> What is the best way to threshold the tracts before sampling FA?
> Firstly, is it best to threshold at the mean group level, or for each individual before calculating the mean.
> Secondly, should I decide on the threshold for each tract separately, or decide on a threshold to use for all of the calculated tracts?
>
> Hope this makes sense
>
> Best,
>
> Pete
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|