Different correlational tractography results with different DSI Studio download versions

[Lisa Marie M.]

Hi,

When doing correlational tractography I am getting very different results depending on which version is used (version downloaded November 22 or May 23). The November version has significant findings (decreased QA correlated to increased Stroop effect) while the May version does not have any significant findings. I have attached parts of the reports for comparison. I am now wondering, what has changed so dramatically in DSI Studio in this period?

Also, can there be different results when running it on Apple OS Ventura, version 13.0, and Windows 10 Enterprise N, version 21H2 (assuming same DSI download date)?

Thanks, in advance, for your help!

Kind regards,

Lisa

[Fang-Cheng Yeh]

The seeding strategy changed, and it affects how ROIs was used in
fiber tracking. The updated version may give better coverage.

The steps you used have one point that caused my attention: the
analysis here used multiple regions which may cause the results to be
unstable. Usually, I would avoid adding many regions of interest
unless I have very strong findings or a very large number of subjects
(e.g. > 500).

Best regards,
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/83a20440-7d0c-467d-9115-3a7786dd83d6n%40googlegroups.com.

[Lisa Marie M.]

Thank you for the explanation!

Just one thing regarding what you mentioned about the regions of interest. We only have one ROI, the others are regions of avoidance which we added because our ROI is a very small structure and we wanted to avoid contamination from larger surrounding areas. You talked about several regions of interest (which we don't have), did you mean the regions of avoidance maybe?
If there is a better way to go about this, I would appreciate any advice!

Kind regards,

Lisa

[Fang-Cheng Yeh]

Using a small ROI is the major problem, because the results will be
very sensitive to misalignment and tracking settings.
I would avoid using ROI unless it is necessary.

You may consider segmenting the whole brain results to get more
specific findings instead of using an ROI.
Frank

[Lisa Marie M.]

For the purpose of the research, the small ROI is necessary but we have everything in MNI space, in case that makes it any better.

But would you generally recommend including ROAs to avoid contamination from surrounding areas, when using a small ROI?

Kind regards, 

Lisa

[Fang-Cheng Yeh]

ROA usually does not matter much. How small was the ROI you used?

> --
> 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/77cd55e6-4370-4e80-b9be-2aa2e32b2a40n%40googlegroups.com.

[Lisa Marie M.]

Here is a screenshot of the regions. ROI is the claustrum, so it's the small structure in between the larger Insula and Putamen (both ROAs)

[Fang-Cheng Yeh]

The acquisition usually does not have enough resolution.

This is a very limiting setting for tractography and is unlikely to produce robust results on correlational tractography.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/99400a41-3b91-474c-a0c6-58810b52e7b7n%40googlegroups.com.

[Lisa Marie M.]

Ok. Thank you for all the advice!