Structural connectivity matrices using HCP data

[Taotao]

Hi Dr. Yeh

First of all, thanks for developing DSI studio and building a supportive DSI community, that makes tractography easier than ever before. I am trying to do some structural connectivity analysis using DSI studio on the HCP data. I realized the correlations between different connectivity matrices from different subjects were pretty low (e.g., the attached Figure), which is alarming. I wonder if something is wrong with my method. I am using the src files you kindly shared in the sample data, the command lines are below:

dsi_studio --action=rec --source=".\100307.src.gz" --method=7 --param0=1.25

dsi_studio --action=trk --source=".\100307.src.gz.odf8.f5.bal.210fx.hcp1021.qsdr.1.25.R78.fib.gz" --track_count=10000000 --output=no_file --connectivity=".\Schaefer2018_100Parcels_7Networks_order_FSLMNI152_1mm.nii.gz" --connectivity_type=pass --min_length=20 --max_length=300 --track_length=20 --connectivity_value=count

I will upload the ROIs to the dropbox if that can help identifying the issue. I do get a warning when I loaded the ROIs into the GUI:

"The nii file has different image dimension. Transformation will be applied to load the region"

Also when I switched to the HCP842_tractography parellation atlas, the correlation seems to be higher. But other atlas in the DSI studio such as AAL2 and  FreeSurferDKT also generate poorly correlated matrices.

Thanks for your time.

Best,

Taotao

[Fang-Cheng Yeh]

Hi Taotao,

    HCP842 Tractography atlas is a white matter atlas and thus is not suitable to be used as the parcellation map for connectivity matrix.

    The correlation won't be highly correlated between subjects because each individual brain is unique in the structural topology. The correlation should be higher in test-retest dataset of the same subject.

    Hope this helps.

Best,

Frank

--
You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/56a321d5-a3ba-452b-be91-0d532986e87b%40googlegroups.com.