Group comparison in Lead Group

[rubenscuryParticipant]

Hi colleagues,
I’m trying to compare variables between 2 groups in the Lead Group (for example, bad vs good responders on motor outcomes in Parkinson’s disease). Patients were settled in the group I or II. When I run the group comparison (two sample t-test) after prepare stats, it appears a screen with “normalized volume intersections, left hemisphere” for example, and “t-value=NaN, p-value=Nan”….I don’t know what is going wrong….Did anybody succeed with group comparison t-test analysis?

Many thanks
Rubens

[andreashornKeymaster]

Hi Rubens,

good to hear from you – could it be that the sample size is too small for a t-test? What is your N of the two groups?
Once you click the button, the raw data should appear in your Matlab workspace as well, so you can easily export and do stats elsewhere, as well.

Hope that helps!

Best, Andy

[rubenscuryParticipant]

Thanks Andy,

I am comparing group 1 (N=20) vs group 2 (N=10)….it still appears the following message in the matlab workspace:
Volume Intersections, both hemispheres

h =

NaN

p =

NaN

ci =

NaN
NaN

and so on….

The correlation is working very well, but the group comparison I am not sure what is happening…

Many thanks
Rubens

[andreashornKeymaster]

Hi Rubens,

maybe your regressor is set up incorrectly. It should be all ones for group 1 and all zeros for group 2.
So assume you have e.g. 5 top responders and 5 poor responders and want to run a paired t-test on their VTA intersections with the STN, then you would add a vector like
[1
1
1
1
1
0
0
0
0
0];
as the regressor. Then click on the group comparison button.

And then it should work.

Best, Andy

[andreashornKeymaster]

(of note, the word “regressor” is wrong here, it has some historic reasons why I’m used to reference the variables entered into lead group as “regressors”). In this case, it’s basically just a variable that codes for the groups.

[rubenscuryParticipant]

Very nice Andy, thanks a lot

I am also trying to change the “brain parcellation” in the Lead Group. For exemple, move from AICHA to Harvard Oxoford…..
In the Lead Group, all the patients and the variables are already set. But when I change in the link “which brain parcellation to use:” from AICHA to Harvard Oxoford and run “Prepare Stats”, in the “fiber count to” still appears the AICHA parcellation (G_Frontal_Sup_Orb…etc…)…i.e. it is not changing the brain area….

Would you have any insight for what is going wrong?

Again, thanks for your attention
Best
Rubens

[andreashornKeymaster]

No it should update like that.
You probably could update to 2.2 and test again?

Best, Andy

[andreashornKeymaster]

Hi Rubens,

I also just noticed that there was a serious bug in the ttest functionality in the “normalized” VTA reports.
https://github.com/netstim/leaddbs/commit/54fdc91f0570d442388f6030f58a56c9337ccf0f#diff-c406b0ce9f31d9fc6834ffef4b9e7f78R917
See this edit – if you need it now would need to change lines 917, 920 and 923 accordingly (or wait for the next update).

Thanks! Andy

[rubenscuryParticipant]

Thanks Andy,

The “brain parcellation” problem I solved, it is working normally…

The “t-test” issue I’ll try later and see….

Best
Rubens

[Author]

Posts