3dReHo with 125 neighbors

AFNI Message Board
1

Hi !

I am wondering whether 3dReHo supports ReHo calculation with 125 neighbors (or 2 extended neighbors each dimension).

Besides, I found the mismatch between the data provided by the help and the that calculated by the given formula “4PI(R^3)/3” with the following list. Which is the corrected volume the program used ?

R V via help V via given formula

2 33 34
2.3 57 51
2.9 93 102
3.1 123 125
3.9 251 248
4.5 381 382
6.1 949 951

Furthermore, can you explain what the unit of the option “-neigh_RAD” is (radius in voxel or in mm) ?

Thanks for the help!

2

Hi, Irene-


I am wondering whether 3dReHo supports ReHo calculation with 125 neighbors (or 2 extended neighbors each dimension).

I assume you mean a 5x5x5 cube centered around a given voxel? No, it doesn’t (at present…). It always made more sense to me to have a spheriodal ellipsoidal shape more than something with corners that is directionally dependent on the coordinate axes.
Is that shape really vital for your work? If not, and the number/volume is important, then I would recommend using ‘-neigh_RAD 3.1’, which would give you a volume of 123 voxels, essentially the same as what you have. If it is vital to have a cube, then an option could be added for that (but, geometrically speaking, rounder shapes would make more sense to me).

Besides, I found the mismatch between the data provided by the help and the that calculated by the given formula “4PI(R^3)/3” with the following list. Which is the corrected volume the program used ?
The mismatch is real and also not an error. The help says that the formula is approximate because the V=4PI(R^3)/3 applies in a continuous context, and the question of voxels is inherently a discretized one. The values given in the help are correct for the present voxel context.


Furthermore, can you explain what the unit of the option "-neigh_RAD" is (radius in voxel or in mm) ?

Values after neigh_RAD and (neigh_X, neigh_Y, neigh_Z) are all given in terms of numbers of voxels. The decimal-y parts affect how the shapes capture voxels that aren’t on the major axes.

–pt

3

Hi ptaylor,

Thanks for your prompt reply. Even though I do agree that the rounder shapes make more sense, the paper stated that two common neighbor sizes are 1 or 2 (corresponding to the cubic box sizes of 3 or 5). Thus, I am wondering whether the cubic option can be add to 3dReHo ?

Ref.
Jiang, L., Zuo, X.N., 2016. Regional Homogeneity: A Multimodal, Multiscale Neuroimaging Marker of the Human Connectome. The Neuroscientist 22, 486–505. doi:10.1177/1073858415595004

Best,

Irene

4

Hi, Irene:

The full quotation from that paper you cite (Jiang and Zuo, 2016) is:
Zang and others (2004) proposed the original ReHo in 3D volume imaging space. The neighbor relationship of a
given voxel was determined by the adjacency of the voxel in either native or standard template 3D spaces (Fig. 1C).
For example, two common neighbor sizes are 1 or 2, corresponding to the cubic box sizes of 3 or 5, which contain
9 or 27 neighbor voxels. We refer to these two ReHo metrics as 3dReHo-1 and 3dReHo-2.

So, I don’t understand how those numbers all fit together, actually. Their Figure 1C does, indeed, appear to show 3x3x3 and 5x5x5 cubes, though.

Also, they do just say that they are “common” neighborhoods-- I don’t know that means those are necessary values.

That all being said, I have added options for ‘-box_RAD …’ and ‘-box_X …’, etc. to allow box shapes; it should be in the next build.

–pt

5

Hi ptaylor,

Thanks for your explanation. I did work and useful.

Irene

6

Rockin’.

–pt