Interpret some interesting restults

[Qiuting Wen]

Dear Gary,

I used NODDI on one brain tumor patient and there are two regions with interesting NODDI values I would love to have you opinions:

1. in some CSF regions, fiso is high (=1) but odi is also very high (=0.704,highest value of the whole brain). In your paper I also notice some bright patches in CSF regions, and do you know what causes high odi in CSF?

2. in one edema region, fiso is moderately high (0.4) - implies there is tissue there. ficvf is kind of low (0.1), odi is very low (0.05), does it mean for the tissue underneath, only a small portion of water signal comes from intra-neurite (sticks model), most of it is extra-neurite (anisotropic gaussian diffusion)? 

Thanks in advance!

Qiuting

[Qiuting Wen]

For 2, I would like to add that in that edema region, kappa is also very high (1~3). Could you help explain what kappa implies?

Thanks!

Qiuting

[Hui Zhang]

Hi Qiuting,

These are very good questions.  My answers are below:

I used NODDI on one brain tumor patient and there are two regions with interesting NODDI values I would love to have you opinions:

1. in some CSF regions, fiso is high (=1) but odi is also very high (=0.704,highest value of the whole brain). In your paper I also notice some bright patches in CSF regions, and do you know what causes high odi in CSF?

It is important to recognise that NODDI uses a hierarchical model.  At the first level, you have either tissue or CSF.

S_total = (1- fiso) * S_tissue + fiso * S_iso

In the event that CSF dominates, i.e., fiso -> 1, (1-fiso) approaches to 0.  Then S_tissue can take arbitrary value to result in the same S_total.

In other words, neurite density and orientation dispersion can take any values.  Put it differently, the estimates in this case, except for fiso, is meaningless.  I hope this makes sense.

2. in one edema region, fiso is moderately high (0.4) - implies there is tissue there. ficvf is kind of low (0.1), odi is very low (0.05), does it mean for the tissue underneath, only a small portion of water signal comes from intra-neurite (sticks model), most of it is extra-neurite (anisotropic gaussian diffusion)? 

Yes, I will interpret this similarly.

Gary

 

[Gary]

kappa and ODI are two ways of looking at the same thing.  kappa is the concentration parameter and ODI is given by kappa.  See the NODDI paper for the definition of ODI in terms of ODI.  Note that, due to some implementation detail, the kappa stored on disk is the true concentration parameter scaled by 0.1.

Gary