Variation in tract vs region stats

[Hannah Lindsey]

Hi! 

I have some longitudinal data that I was taking a closer look at and in doing so, I created a region from a tract for a single subject, exported stats from both the region and the tract, and then compared them. I wasn't expecting them to be exactly the same, but the average values are pretty different across metrics and pattern of the DTI metrics is almost opposite between the two stat exports for MD, AD, and RD (patterns are more similar for GQI metrics).

In the attached screenshot, the graphs in the top row show region stats, and those in the bottom are the tract stats. I've also attached a table of the data, and the command history log so you can see that I did all the steps at once, so there's no variability in registration or anything before extracting the stats.

Can you help me understand how this happened?

Thanks!

Hannah

[Frank Yeh]

If the region is near the border of the mask, the zero background may largely reduce values during interpolation.

Only along-tract values were estimated through interpolation, and thus if tract values are lower, then this is the most likely cause.

Therefore, we will only compare values between tracts and tracts, regions and regions.

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/3ed0f64d-90c2-4692-bfc8-f27ccca4b2fan%40googlegroups.com.

[Hannah Lindsey]

Oh duh… I should have realized that. This was the corpus callosum, so really close to the ventricles. That all makes sense. Thank you!

_________________________________

Hannah M. Lindsey, PhD

Research Associate

TBI and Concussion Center

Department of Neurology

University of Utah

On Jun 20, 2025, at 8:35 am, Frank Yeh <fran...@gmail.com> wrote:



[Frank Yeh]

It is pretty hard to figure out (unless you write the code).
Frank