[Mristudio-users] CC in elderly patients

[Jun Yi Wang]

Hi Susumu,

Thank you very much for the detailed explanation!  It is very useful.  I have the same impression of the difficulty of defining a gold standard for tractography after I read the paper by Dauguet et al. (2007): Comparison of fiber tracts derived from in-vivo DTI tractography with 3D histological neural tract tracer reconstruction in a macaque brain.

Jun Yi




________________________________
From: susumu mori <susumu at mri.jhu.edu>
To: "DTI Studio, ROI Editor, Landmarker Questions/Support" <mristudio-users at mristudio.org>; Xin Li <xli16 at jhmi.edu>
Sent: Thu, March 11, 2010 12:01:50 PM
Subject: Re: [Mristudio-users] CC in elderly patients

Hi all,

We have noticed the same phenomena and also got several inquiries from outside. Here is my understanding;

1) For young subjects, the shape of the CC has a clean U shape in coronal sections. If you do fiber tracking, you usually can reconstruct the U-shape trajectory, which makes some anatomical sense.
2) For the population with enlarged ventricles, such as elderly, AD, and stroke patients, the trajectory of the CC becomes more like a "W" shape with a sharp turn where the CC merges to the corona radiata. I believe that this is the primary reason that most streamlines end up in going downward toward the internal capsule, which doesn't make much anatomical sense. 
3) Think about the 90 degree rotated "T" shape and the vertical line is the corona radiata and the horizontal line is the CC. If the horizontal line merges at the right angle, the probability of the tracking to go upward and downward is 50/50. If the horizontal line is slightly tiled upward, the CC tracking merges to the vertical line (corona radiata) and continues to travel upward. It's as simple as this. Once we get a vector (or tensor) field and ask a computer to reconstruct a path, the computer doesn't know the anatomy. All it knows is just a bunch of vectors. One may think that probabilistic tracking can deal with it but the connectivity strength could be influenced by the merging angle too.
4) This exemplifies what tractography is. It is just a region growing (or pixel-connection) tool based on pixel-by-pixel information, no matter which algorithm we use.  It is just a feature extraction tool from MRI that observes water molecules. Some results look right and some results do not.
5) Then many people ask, "how do we know the validity". Here we (both who ask the question and who is asked the question) have to think very hard. To show the validity, we need to compare our observation (in this case tractography result) and gold standard (real anatomy). Then what do we really mean by "real anatomy" or "real connectivity"? What is connectivity anyway? First of all many of the macroscopic anatomical definition such as "superior longitudinal fasciculs" and "corona radiata" are defined by human eyes based on anatomical features, which doesn't have a clear definition and boundary. It's like a cloud. Everybody can see it at the core but at the boundary, we don't know how to define. Then how about the microscopic anatomical definition? Yes, we have a very clear definition, which is axonal connection. However, we know that one neuron communicate with multiple neurons through a branching axon. We have 100 billion neurons. Even if one neuron
 communicate with one neuron, this is a 100 billion x 100 billion question. For MRI, even if you do 1,000-orientation Q-ball imaging, we are talking about merely few GB of data. With a tensor fitting, what we have is 20 MB. On top of it, we are measuring water properties. So, we can understand that we are on a very shaky ground when we talk about "validation" because we can't even define what is the gold standard we are shooting for and our data is so small compared to what we are interested in.
6) Then next question is, "why do we do tractography". Here, we should go back and think why we do MRI first of all and what is the real power of it. The real power is that we can examine the entire brain within 30 min non-invasively and quantitatively. This is something no other modality, certainly histology, can't do. By performing a feature extraction such as tractography, we can quickly and quantitatively assess the anatomical status.
7) As long as we apply the same data acquisition method and the same analysis method, the results itself is always correct. There is nothing wrong with the results whatever they are. In this example, in some population, the CC goes downward while it goes upward in the normal subjects. The tractography sensitively detected some differences between the two population.
8) If we make mistake in this situation, that is when we interpret the data. For example, if we interpret the data like, "in young brains, the CC goes up and in aged brain the CC goes down because the connectivity changed", now such a statement becomes the subject of "right or wrong".
9) So, it is very important to evaluate why you got different results between two groups. The real reason could be something not the measurement was intended. For the case of the CC trajectory, the real reason could be "enlarged ventricles" and the DTI + tractography could be just an expensive and indirect way to show it.
10) This is a very typical issue in MRI, not just DTI. When we observe cortical thinning in T1, it may be just a change in contrast due to myelination status or other things. What we really observe is the thinning of T1-indicated-cortical-compartment and the direct translation into the real cortical thickness is sometimes dangerous or even erroneous. As long as we are dealing with 10 MB of water information, biological or cellular interpretation always requires caution.
11) Fiber reconstruction is a high-level image analysis tool, which is sensitive to so many anatomical factors including both micro and macroscopic factors. The microscopic factors are myelination, axonal loss, cellular loss, edema, etc that would affect the FA and anatomy. The microscopic factors are multiple compartments (e.g. crossing fibers) within the 2 mm pixel. So, in general, the interpretation would be difficult. It is a good tool to survey the overall white matter anatomy but you need to exercise the caution when you interpret it.


I guess what I mentioned above contains a lot of my personal opinions and please take it as just one of the opinions, but hope it helps to understand the issue of the downard CC projection.

Susumu



On Thu, Mar 11, 2010 at 11:52 AM, Jun Yi Wang <junyiwang2002 at yahoo.com> wrote:

Hi Dafna,
>
>I only have a few subjects in 80s, but I have subjects from 16-65 years old.  I do see a substantial decrease of the amount of corpus callosum fibers projecting towards the cortex as we age.  Some fibers still project to the cortex although a lot of them project downwards instead.  I don't know whether this is caused by aging or DTI artifacts such as the termination of line propagation because of the lower FA or the crossing fiber issue.
>
>Jun Yi Wang
>
>Postdoctoral
> Research Scholar
>Fragile X Research and Treatment Center
>UC Davis M.I.N.D. Institute
>530.747.3808 (Lab)
>
>
>
________________________________
From: Dafna Ben Bashat <dafnab at tasmc.health.gov.il>
>To: mristudio-users at mristudio.org
>Sent: Thu, March 11, 2010 3:19:20 AM
>Subject: [Mristudio-users] CC in elderly patients
>
>
>
>>
>Hi
>all,
>We
>are currently working on data from elderly stroke/TIA patients.  
>In
>many of the TIA patients, who have good DTI data, for some reason the CC does
>not built well, we can't see the fan. 
>This
>is not due to incorrect directions (the CST is built OK).
>ROI
>measurements give reasonable values. We even changed the flip-angel for the
>FACT, and got the same results.
>Any
>suggestions? Is it a known phenomenon in elderly / stroke / TIA patients? 
> 
>Thank
>you in advance.
> 
>Dafna
>
> 
> 
>Dafna Ben Bashat Ph.D. 
> 
>Senior Physicist, In charge of MR systems
>The Wohl institute for Advanced Imaging
>Brain Imaging Center
>Tel Aviv Sourasky Medical Center
> 
>Phone: 972-3-6973953
>(o)          Fax: 972-3-6973080
> 
>       972-52-4262515
>(m)        Mail: dafnab at tasmc.health.gov.il
> 
>
>_______________________________________________
>>Mristudio-users mailing list
>Mristudio-users at mristudio.org
>http://lists.mristudio.org/mailman/listinfo/
>>Unsubscribe, send a blank email to: Mristudio-users-unsubscribe at mristudio.org
>
>



      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20100311/d4519ecc/attachment-0001.html