Hi Longchuan,
>extremely helpful
:)
>Do you think waytotal is biasing towards fibers that have large diameters or higher coherence?
To be honest, I have no idea and I would not know what to use as a reasonable reference in such comparison.
I don't see why "coherence" would increase waytotal numbers.
>better ideas to give those connection strength between smaller cortical areas more weights
Not off the top of my head. Given that the waytotal increases exponentially the more seed voxels are connected to the target (see Figure 19.11 in the chapter), that's no easy thing.
Cheers-
Andreas
PS: Sorry if this is not so helpful...
________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von Long Chuan [[log in to unmask]]
Gesendet: Donnerstag, 17. Dezember 2009 19:54
An: [log in to unmask]
Betreff: Re: [FSL] AW: [FSL] interpretaion of waytotal number
Hi Andreas
Thank you very much for your comments. They are extremely helpful. I will check out your study relating the waytotal with the motor functions.
I have a few follow-up questions and would really appreciate if you would help on that.
The reason for me to compare tracts with different strength (waytotal) is that I am constructing a brain network based on the connectivity strength derived from Fdt toolbox. i.e., with different brain regions as nodes and the normalized waytotal number connecting each two cortical region pairs as links. One advantage that I think that probability tractography can give us is that we can threshold the connectivity matrix based on these values to get rid of those spurious connections that will exist in the network either built based on diffusion tensor imaging or diffusion spectrum imaging techniques. However, I do notice that projection and commissure fibers usually have higher connectivity strength than that in the association fibers as reflected in the waytotal. Do you think waytotal is biasing towards fibers that have large diameters or higher coherence?
My second question is about the normalizing the waytotal by the areas of the cortical masks. It seems to me that when the fiber density on cortical masks is not even, normalizing the connectivity based on the cortical areas is not a very justified way to do, even it has been used in some papers. Do you have any better ideas to give those connection strength between smaller cortical areas more weights?
Thank you very much in advance!
Longchuan
________________________________
From: Andreas Bartsch <[log in to unmask]>
To: [log in to unmask]
Sent: Wed, December 16, 2009 4:51:19 PM
Subject: [FSL] AW: [FSL] interpretaion of waytotal number
Hi Longchuan,
>two cortical target masks, one at the thalamus and another at precentral gyrus
the thalamus is not cortical.
W/r to your hypothetical examples: I would never compare the two cases. You should essentially i) have knowledge that the tract exists (=connects your masks) in every case you examine. Knowledge can be clinical. ii) compare similar tracts iii) think hard if / how you want to adjust for different masks sizes.
The waytotal was derived to obtain the probability that a voxel v is part of the tract between A and B given that we know that A<->B exists. In this cases, you can simply devide the number of samples passing through v (given by the intensity of v in the fdt_path output of probtrack(x)) by the waytotal. It's simple Bayesian maths..
To some extent, the waytotal does reflect pathway strength. However, the question is what is a good reference, what to compare with.
>studies that have correlated the waytotal number with anatomical pathway strength or any functional significance
Yes: check out our chapter on Tractography for surgical targeting in Heidi's & Tim's book. Figure 19.12 may be of interest. Here, we used the same mask sizes for homologous seed/target masks across the hemispheres of each subject but not across subjects. Note that in the extreme the waytotal of the CST portion tracked was up to 5 times higher in the lesional hemisphere. Still: the fewer the number of samples that reach the target from the seed relative to the contralateral side, the more pronounced the motor weakness in our sample. The waytotal reduction compared to the nonlesioned side was in better concordance with the clinical rating of the motor weakness than the FA reductions. So it can be useful but requires caution.
Cheers-
Andreas
________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]<mailto:[log in to unmask]>] im Auftrag von Long Chuan [[log in to unmask]<mailto:[log in to unmask]>]
Gesendet: Mittwoch, 16. Dezember 2009 22:19
An: [log in to unmask]<mailto:[log in to unmask]>
Betreff: [FSL] interpretaion of waytotal number
Dear FSL experts
I have a few questions about the interpretation of the waytotal number and would really appreciate if someone would help me on this.
My question is: are there any studies that have correlated the waytotal number with anatomical pathway strength or any functional significance based on the underlying pathway from which the waytotal is derived?
In other words, is it safe to use waytotal number as an index of the anatomical pathway strength?
My previous experience tracking on corticospinal tract kind of says no, because it was quite often to see that waytotal number in the left CST was 3 (or even higher) times larger than that in the right CST in one subject and vice versa in another subject and we could not correlate handedness measures with the waytotal numbers .
On the other hand, it seems that it does reflect a certain degrees of connectivity strength. One example is as follows:
Consider the following two situations:
Case1: we have two cortical target masks, one at the thalamus and another at precentral gyrus, among which a strong pathway is supposed to exist between (CST pathway). We run probtrackx using the whole brain white matter as the seed mask and two cortical masks as the target masks ( 1000 samples per voxel).
Case2: we have two cortical target masks, one at the superior frontal cortex and another at the occipital cortex (among which a relatively weak connection exists) and again the whole brain white matter mask is used as the seed mask and two cortical masks are used as the target masks.(1000 samples per voxel are sent).
A very possible result after running probtrackx on the above two cases is that we get higher waytotal number in the case 1 than that in the case 2 on a population level, even when the areas of the cortical target masks are considered in the analysis. This should also probably reflect some real white matter anatomical differences in the brain as we do see that axons at the CST are much more coherent and their diameters are much larger.
Anyway, I would really appreciate if anyone would give any thoughts or comments on this question
Thank you all in advance!
Longchuan
|