Re: [dsi-studio] Questions about connectometry in brain-injured patients
22 views
Skip to first unread message
Frank Yeh
Jul 5, 2025, 12:34:33 PM7/5/25
to yuw...@ucsd.edu, DSI Studio
I am sorry for the delayed reply.
On Tue, Jun 24, 2025 at 10:34 PM 'Yusheng Wang' via DSI Studio
<dsi-s...@googlegroups.com> wrote:
>
> Hi Frank,
> 1. On the conceptual level:
> If I understand correctly, connectometry performs a voxel-wise correlation between a diffusion metric (e.g., QA, FA) and a behavioral score,
Yes
> and then reconstructs tracts passing through the significant voxels.
Not exactly. It tracks the exact part of tracts based on the
T-statistics threshold.
You may check out the tutorial videos at practicum.labsolver.org about
how correlational tractography works.
>
> However, if the hypothesis is about disconnection between specific brain regions, wouldn't any lesion along the connecting tract lead to similar behavioral consequences?
Not really. For mostly not the case, the tract degeneration is not
homogeneous, and the relation with behavioral scores are also
nonlinear.
> That is, the functional impact is not necessarily localized to a specific voxel but can occur anywhere along the pathway.
My experience is that it is always localized to specific regions.
Rarely it is along the entire pathway.
>
> Would this make connectometry less sensitive to detecting distributed disconnection effects in such scenarios?
The opposite.
Because of the fact that the correlations are always limited to a
segment, it will have much higher power than tractometry.
>
> 2. On the tractography step:
> After identifying significant voxels, I understand that connectometry reconstructs tracks passing through those regions. My question is: in lesion patients, these regions may be structurally damaged and may no longer support valid tractography.
Not really.
In correlational tractography, the local fiber orientations are from
the normal population template, and the T-statistics are from the
data. Structural damage will be captured by T-statistics, and won't
affect local fiber orientations.
> In such cases, does connectometry perform tractography using each patient's own data (which may be impaired),
No,
> does it reference normative tractography (e.g., from control subjects or a template) to reconstruct the tracks through those significant voxels?
Yes
Hope this helps clarify how cT works.
Best regards,
Frank
On Tue, Jun 24, 2025 at 10:34 PM 'Yusheng Wang' via DSI Studio
<dsi-s...@googlegroups.com> wrote:
>
> Hi Frank,
> 1. On the conceptual level:
> If I understand correctly, connectometry performs a voxel-wise correlation between a diffusion metric (e.g., QA, FA) and a behavioral score,
Yes
> and then reconstructs tracts passing through the significant voxels.
Not exactly. It tracks the exact part of tracts based on the
T-statistics threshold.
You may check out the tutorial videos at practicum.labsolver.org about
how correlational tractography works.
>
> However, if the hypothesis is about disconnection between specific brain regions, wouldn't any lesion along the connecting tract lead to similar behavioral consequences?
Not really. For mostly not the case, the tract degeneration is not
homogeneous, and the relation with behavioral scores are also
nonlinear.
> That is, the functional impact is not necessarily localized to a specific voxel but can occur anywhere along the pathway.
My experience is that it is always localized to specific regions.
Rarely it is along the entire pathway.
>
> Would this make connectometry less sensitive to detecting distributed disconnection effects in such scenarios?
The opposite.
Because of the fact that the correlations are always limited to a
segment, it will have much higher power than tractometry.
>
> 2. On the tractography step:
> After identifying significant voxels, I understand that connectometry reconstructs tracks passing through those regions. My question is: in lesion patients, these regions may be structurally damaged and may no longer support valid tractography.
Not really.
In correlational tractography, the local fiber orientations are from
the normal population template, and the T-statistics are from the
data. Structural damage will be captured by T-statistics, and won't
affect local fiber orientations.
> In such cases, does connectometry perform tractography using each patient's own data (which may be impaired),
No,
> does it reference normative tractography (e.g., from control subjects or a template) to reconstruct the tracks through those significant voxels?
Yes
Hope this helps clarify how cT works.
Best regards,
Frank
Reply all
Reply to author
Forward
0 new messages