How to definite specific white matter tracts anatomically

[fullcream]

Hi experts:

The whole brain white matter fiber bundles are tightly packed together, without obvious boundaries to seperate them. But how do we parcellate white matter tracts like IFOF, ILF, etc. from whole brain white matter? I refer to anatomical rather than MR tractography. In other words, how to definite specific white matter tracts anatomically?

Thank you
Gang.

[Steven]

Hello,

@fullcream

I refer to anatomical rather than MR tractography.

I am sorry but I do not understand the difference here.

There is no universal way of defining white matter tracts. Some of them will be ROI based (e.g. AFQ) - that is, a seed and target region for the tract are defined in the subject’s space, and all streamlines that connect the seed and target make up the tract. Some are streamline based (e.g. RecoBundles) - subject tracts are defined based on a model set of tracts. Others use neural nets (e.g., TractSeg) to predict tract streamlines. All have their pros and cons.

Best,
Steven

[fullcream]

Thank you for your patient answer! It solved my confusion!

So we have many ways to define white matter tracts in MR tractography.
there is no universal way of defining white matter tracts in anatomical?

I have another related question:
I want to get connectivity between two ROIs. Is it equivalent for the following two methods:

1. Perform whole brain tractograpy and parcellate it using the criterion that streamlines passing two ROIs.
2. Set the two ROI as seed points and then perform tractography.

PS: Please ignore the difference between “passing” or “seeding”

Best,
Gang.

[Steven]

@fullcream

So we have many ways to define white matter tracts in MR tractography.
there is no universal way of defining white matter tracts in anatomical?

I still am not quite sure what you mean by this, but if you are asking if we can use a T1w image to define tracts, then the answer is no. We use diffusion imaging because it allows us to look at the direction of diffusion, from which we can infer the existence and orientation of white matter bundles.

@fullcream

I want to get connectivity between two ROIs. Is it equivalent for the following two methods:

1. Perform whole brain tractograpy and parcellate it using the criterion that streamlines passing two ROIs.
2. Set the two ROI as seed points and then perform tractography.

These methods would likely not be equivalent. For option 2, you can explicitly state the number of attempts to make streamlines between the seed and target, so that would bias your streamline count. That being said, as long as you choose one method and apply the same across subjects, that should be fine.

How are you defining connectivity? Pure streamline counts are being phased out in favor of streamline weighting informed by underlying microstructure (e.g. SIFT2, COMMIT, LiFE models).

Best,
Steven

[fullcream]

@Steven

I still am not quite sure what you mean by this, but if you are asking if we can use a T1w image to define tracts, then the answer is no. We use diffusion imaging because it allows us to look at the direction of diffusion, from which we can infer the existence and orientation of white matter bundles.

I mean is there a specific anatomical definition of white matter bundles (eg. IFOF) based on anatomical methods (eg. tract-tracing) rather than DWI methods?

@Steven

How are you defining connectivity? Pure streamline counts are being phased out in favor of streamline weighting informed by underlying microstructure (e.g. SIFT2, COMMIT, LiFE models).

Thank you for your reminder, we are considering using LiFE instead of NOS.

[Steven]

@fullcream

I mean is there a specific anatomical definition of white matter bundles (eg. IFOF) based on anatomical methods (eg. tract-tracing) rather than DWI methods?

Tract-tracing is a DWI method.

@fullcream

Thank you for your reminder, we are considering using LiFE instead of NOS.

I would personally recommend SIFT2 or COMMIT over LiFE, but any of these should be more biologically related than NOS.

Best,
Steven

[fullcream]

@Steven

I would personally recommend SIFT2 or COMMIT over LiFE, but any of these should be more biologically related than NOS.

Ok, I plan to learn about SIFT2 and COMMIT next.