Connectivity matrix results questions

[Борщик Борщевой]

Hello!

I've been using [Step T3: Fiber Tracking & Visualization] setting different parameters for each fiber tracking on the same FIB file. And I noticed several things that I wanted to ask about:

1. When I set different Min Length for tracking with keeping other parameters same (
Tracking Index = qa; Tracking Threshold = 0,001; Angular Threshold = 60; Step Size = 0,5 mm; Max Length = 400 mm; Terminate if 1000000 tracts; Smoothing = 0,5)   , I get strange results: Min Length supposed to remove tracks shorter than set parameter, but when I set min Length 5 mm results in connectivity matrix shows less tracks passing regions than when I set min length 20 mm. What could be the reason behind increasing of connectivity with min length growth?

2. While studying connectivity matrix obtained with different tracking parameters I also noticed that no matter what parameters I set or what FIB file I use minimum connectivity (non zero) is always 1000 times smaller than maximum connectivity in same matrix. Why is it this way?

[Frank Yeh]

> Tracking Index = qa; Tracking Threshold = 0,001; Angular Threshold = 60; Step Size = 0,5 mm; Max Length = 400 mm; Terminate if 1000000 tracts; Smoothing = 0,5) , I get strange results: Min Length supposed to remove tracks shorter than set parameter, but when I set min Length 5 mm results in connectivity matrix shows less tracks passing regions than when I set min length 20 mm. What could be the reason behind increasing of connectivity with min length growth?

There will be fewer tracts reaching the cortical regions.

>
> 2. While studying connectivity matrix obtained with different tracking parameters I also noticed that no matter what parameters I set or what FIB file I use minimum connectivity (non zero) is always 1000 times smaller than maximum connectivity in same matrix. Why is it this way?

Because of connectivity threshold that zeros the matrix if the value
is smaller than 0.001

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/88c54959-70f7-463d-832f-40deec3c9d97n%40googlegroups.com.

[Борщик Борщевой]

Thank you for your respond, but I still got some questions regards topic:

1. If we take tracking in range 20 mm - 400 mm it supposed to be less connectivity than in 5 mm - 400 mm range since 5 mm - 400 mm includes 5 mm - 20mm tracks. Is it happens because of autotrack tolerance (== 24 mm) or just DSI studio removes very short tracks from final connectivity matrix as possible false tracks?
2. Also wanted to ask about possibility of usage of weighted graphs to analyze results in connectivity matrix since in all data I got there always around 20 tracks (attached fig) which have very strong connectivity and make all graphs look the same? Why there such standing out tracks and would you maybe recommend some other way to analyze connectivity?

пятница, 1 сентября 2023 г. в 18:30:21 UTC+3, Frank Yeh:

[Frank Yeh]

On Wed, Sep 6, 2023 at 12:50 PM Борщик Борщевой <borschik...@gmail.com> wrote:

Thank you for your respond, but I still got some questions regards topic:

1. If we take tracking in range 20 mm - 400 mm it supposed to be less connectivity than in 5 mm - 400 mm range since 5 mm - 400 mm includes 5 mm - 20mm tracks. Is it happens because of autotrack tolerance (== 24 mm) or just DSI studio removes very short tracks from final connectivity matrix as possible false tracks?

It is always harder for shorter tracts to connect regions. I will say 20mm would have a better chance of being counted in the connectivity matrix.

You may check the mean length of the connectivity matrix and will find mostly would have a mean length much longer than 5 mm.

 

2. Also wanted to ask about possibility of usage of weighted graphs to analyze results in connectivity matrix since in all data I got there always around 20 tracks (attached fig) which have very strong connectivity and make all graphs look the same? Why there such standing out tracks and would you maybe recommend some other way to analyze connectivity?

I always recommend NOT to use a connectivity matrix derived from whole brain fiber tracking, because it is not reliable.

Differential tractography and correlational tractography would give much more reproducible results: practicum.labsolver.org

Best,

Frank

[Борщик Борщевой]

Hello, Frank. Thank you for response.
Now I have another question: I use 0,5 mm step size with 2 mm voxels. Brain has around 240 000 voxels in it and each voxel has 4-12 tracts passing it with such step size. So should I use 3 million as number to terminate tracking or it will create more false connections?

среда, 6 сентября 2023 г. в 20:02:46 UTC+3, Frank Yeh:

[Frank Yeh]

The ratio of false connections will be mostly the same.
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/9b4e88c2-8540-470b-895a-b099200bff56n%40googlegroups.com.