Hi Matt,

It's interesting to discuss about the connection between BA6 and temporal lobe, since recently I saw some papers claiming the arcuate fasciculus actually ends in BA6, but not the classical Broca's.
I don't know if you have read this review which touched this issue: http://brain.oxfordjournals.org/content/132/9/2309.abstract

For the crossing fiber problem, could you illustrate a bit what you meant by  " you might in fact be measuring a change in the arcuate (...) or SLF III.. "? 

And the following-up question would be: does the probabilistic tracking handle the crossing fibers so badly that I should not trust at all the large amount of fibers I got between BA6 and temporal lobe?

Best,
hd







On 16 May 2011 22:47, Matt Glasser <[log in to unmask]> wrote:

I don’t believe there is a real connection between BA 6 and posterior temporal cortex, as I have looked at higher and higher quality diffusion data (better resolution, more directions, higher bvalue) and it becomes less and less, as one does a better job of resolving the crossing fibers.  Granted these are my opinions, so your mileage may vary. J  To be convinced I would have to see a strong suggestion in the data that you could distinguish SLF III connections and arcuate connections in this region and that you still found arcuate connections of higher probability than SLF III connections.  The issue is that you might in fact be measuring a change in the arcuate (which I think goes to BA 44, 45, 47, and DLPFC) or SLF III (which I think goes to BA 6 and 44) and attributing this difference to a pathway that may not exist.   

 

As for your other question, it has been published before:

 

Hua, K., J. Y. Zhang, S. Wakana, H. Y. Jiang, X. Li, D. S. Reich, P. A. Calabresi, J. J. Pekar, P. C. M. van Zijl

and S. Mori (2008). "Tract probability maps in stereotaxic spaces: Analyses of white matter

anatomy and tract-specific quantification." Neuroimage 39(1): 336-347.

 

But they weren’t the only ones to have that idea…

 

Peace,


Matt.

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of hd x
Sent: Monday, May 16, 2011 3:44 AM


To: [log in to unmask]
Subject: Re: [FSL] Highly negatively correlated waytotal and average FA

 

Hi Matt.

 

Thanks a lot for these hints!

 

I never heard about using "a weighted (by probability) average of FA" before. But it sounds very delicious. Could you recommend any reference on this method?

 

As for the BA6-posterior temporal connection, our hypothesis doesn't care which one or two fiber(s) it includes, as long as there are real fibers exist. While the concern is that if the significant crossing fibers will weaken the interpretation of the correlation between the behavior and the waytotal?

 

Best,

hd 

 

On 13 May 2011 20:05, Matt Glasser <[log in to unmask]> wrote:

You could take a weighted (by probability) average of FA inside the unthresholded pathway to avoid the possible interaction you mention.  As for the pathway you are attempting to track, there are significant crossing fiber issues there and I have my doubts that there really is an arcuate connection to BA6 vs a confusion between the arcuate and SLF III.

 

Peace,

 

Matt.

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of hd x
Sent: Friday, May 13, 2011 3:50 AM


To: [log in to unmask]
Subject: Re: [FSL] Highly negatively correlated waytotal and average FA

 

Hi Matt., I traced the pathway between BA6 and posterior temporal lobe. Thus it goes along the arcuate fasciculus.

 

I suspect the corr. between FA and waytotal is caused by the low thresholding. Pathways with larger waytotal numbers have more “marginal” voxels survived after thresholding, which leads to smaller average FA. 

How do you think?

On 12 May 2011 16:30, Matt Glasser <[log in to unmask]> wrote:

Okay, I understand what you did.  Are you tracking a pathway along a non-primary fiber orientation?  That could cause a negative correlation between waytotal and FA (i.e. the stronger the pathway, the lower the FA because of crossing effects).  It would help to know the pathway as well, but if you don’t want to publically say, you could tell me off list.

 

Peace,

 

Matt.

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of hd x
Sent: Thursday, May 12, 2011 1:49 AM

Subject: Re: [FSL] Highly negatively correlated waytotal and average FA

 

Hi Matt.,

 

I made an error in describing how I normalized. I didn't divide the waytotal, but divided the number of streamlines per voxel along the pathway.  But in case you got what I meant, are you suggesting to take the average FA on a not-thresholded pathway?

 

Best,

HD

 

 

On 11 May 2011 18:33, Matt Glasser <[log in to unmask]> wrote:

What happens if you don’t do this normalization.

 

Peace,


Matt.

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of hd x
Sent: Wednesday, May 11, 2011 3:53 AM
To: [log in to unmask]
Subject: [FSL] Highly negatively correlated waytotal and average FA

 

Dear list,

 

I'm working on a study examining the relationship between a behavioral measure and a particular white matter pathway defined using probtrackx (part of the arcuate fasciculus), and I found:

1. The behavioral measure is negatively correlated with the waytotal for the pathway but positively correlated with average FA along the pathway.


2. There is a strong Negative correlation between waytotal and average FA along the pathway.

 

The above two points seem to suggest something circular. Would this be due to a problem of low thresholding or something related to the probabilistic algorithm?

 

Ps, I divided the waytotal by the total sample sent out to normalize the pathway and took a threshold of 5X10-6 .

 

Any thoughts or experience are highly appreciated.

 

Best regards,

 

hd