Connectivity Matrix Resutls versus Filter Tracts by ROI Results

[Allison Yan]

Hi Frank,

I was using the Connectivity Matrix function in DSI Studio and wanted to verify my matrix results. I used the same parameters for whole-brain tracking and used the HCP-MMP parcels as the ROIs. I found some differences between my matrix results and the filter tracts by ROIs function. 

For example, in my matrix, the streamline count between 47l and V1 is 1359, which exactly matches the result from the “filter tracts by ROI” function. However, the streamline count between a10p and V1 is 0 in the matrix, while the “filter tracts by ROI” function shows 27 streamlines. I also tried checking other connections between different parcels. So far, it seems that connections with fewer than 30 streamlines in the “filter tracts by ROI” function sometimes appear as 0 in the matrix. 

Below is the screenshot of how I calculated the matrix with the HCP-MMP atlas.

I’m wondering whether my approach is correct. If so, would this be an issue when interpreting the matrix results, or is there another solution you would suggest?

Thank you so much for your help!

[Frank Yeh]

There is a checkbox named "threshold by" with a value of 0.001, which zeros entries with values lower than that threshold. Setting it to zero to see if it matches.

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 visit https://groups.google.com/d/msgid/dsi-studio/a937d6c7-5f0b-4fd9-88e6-40353c4d5683n%40googlegroups.com.

[Allison Yan]

Hi Frank,

Thanks for your faster response! Now it matches! I am wondering whether you have any suggestions about how to set the threshold. 

Best,

Allison

Frank Yeh <fran...@gmail.com> 于2026年6月10日周三 15:17写道：

[Frank Yeh]

Hi Allison,

For the threshold, it really depends on your analysis goal. If you want to include all potentially existing connections, set the threshold to 0 (or the minimum possible value). If you are focusing on robust connections, you might set it based on a minimum number of streamlines or a relative percentage of the total streamlines found.

For now, setting it to 0 or a very small number like 0.0001 is usually safe if you want to see everything.

Best,
Frank

[Allison Yan]

Hi Frank,

Got it! Thank you again for all the explanations.

Best,

Allison

Frank Yeh <fran...@gmail.com> 于2026年6月10日周三 15:35写道：