Connectome stats


#1

Hi,

I’ve for two quick questions re structural connectome stats:

I’m currently trying to run some group comparisons for a structural connectome in NBS connectome. The stats work well, but I’m having trouble with the visualization. Does someone have the coordinates for fs_default.txt in MNI space? When I run getCOG.m with my parcellated image, it seems to rotate the images by 90 degrees and confuse axial, saggittal and coronal when overlaying on the figure.

I followed the instructions on the MRtrix website for the structural connectome reconstruction, so I guess my current martices are populated with the streamline count. Is this correct? How can I populate the matrices with FA and other scalars?

Cheers, Lena


#2

Hey @Lena13,

As noted in the email, Ill reply with the most simple for now:

Yes, the default is to calculate the streamline counts. To calculate the mean FA over the structural connection, should be to add the following arguments to tck2connectome:

-scale_file yourfaimage.nii -stat_edge mean

As noted in email correspondence, the apparent fiber density is a bit more tricky, as the heuristic (http://mrtrix.readthedocs.io/en/latest/reference/commands/afdconnectivity.html) here is dependent on:

  • All images are required to be intensity normalized
  • Sum of FOD amplitudes are corrected by the streamline length, and
  • Either calculated for fixel with greatest density of streamlines, or all fixel voxels (i.e. -all_fixels argument).

The issue is that perhaps your subject’s parcellation is in native space? And hence you would need to co-register your image into MNI space.

With that being said, I would first check the strides of your parcellation image?

Best,

Alistair


#3

Hi Alistair,

Thanks this is very helpful. I managed to figure out the coordinate file. However, when I try to run the tck2connectome script with my FA image, I get the following error, not matter if I run if with a .mif or .nii file:

[ERROR] File “P1001_fa.nii” contains non-numerical data

Any thoughts on what the problem might be?

Cheers, Lena


#4

Hi @Lena13,

I’m receiving the same issue. I figured it would be the option for creating FA connectomes.

@rsmith or somebody else may be able to help with this.

Regards,

Alistair


#5

My current martices are populated with the streamline count. Is this correct? How can I populate the matrices with FA and other scalars?

Yes, the default is to calculate the streamline counts. To calculate the mean FA over the structural connection, should be to add the following arguments to tck2connectome:
-scale_file yourfaimage.nii -stat_edge mean

Not quite; There’s an additional explicit step required, as documented here: you need to first calculate the mean FA per streamline and store these values in a text file, then feed this to tck2connectome, where the mean “mean FA” per edge is calculated. That’s why you’re getting this error:

[ERROR] File “P1001_fa.nii” contains non-numerical data

… because the -scale_file option expects numerical data :stuck_out_tongue:


#6

That was working perfectly. Thank you! :slight_smile: