-gltsym advice

AFNI Message Board
1

Hello,

I ran a block design where participants are wither moving their Left Hand, Right Hand or Both Hands. One question I would like to answer is if Both Hand movement activation is greater than Left Hand and Right Hand activation and does Both Hand movement activation result in unique regional activation.

I was wondering if you could offer advice on how to set up the -gltsym

What syntax would I use to set up?

(Both - Left) + (Both -Right)

Thanks

2

(Both - Left) + (Both -Right)

Your setting would show brain regions with statistical evidence for Both being different from the average of Left and Right. Alternatively,

-gltsym ‘SYM: +Both -Left \ +Both -RIght’

shows brain regions with statistical evidence for Both being different from eft or Right.

3

Thank you for your response.

How would I find where Both is different from Left AND Right?

You’re right, I do not want to the first expression because I do not want to average Left and Right because that might influence the emphasis of one vs. the other. The second expression will be difficult to decipher the uniqueness of both, because I will always find significance in the hemisphere ipsilateral to the moving hand just because of the contralateral motor activation inherent to moving the left hand and the right hand.

Thanks again.

4

Perform two separate inferences:

-gltsym ‘SYM: +Both -Left’

and

-gltsym ‘SYM: +Both -RIght’

Then use the intersection (conjunction) of the two as the evidence for Both being stronger than both Left and Right. However, be careful in interpreting the result: It does not mean that region outside of the conjunction is the opposite: “lack of strong evidence for effect X” does not mean “strong evidence for the lack of effect X”.

1 Like
5

Thank you!

For the conjunction analysis would you recommend:

3dcalc -prefix conjunction_map -a contrast1 -b contrast2 -expr ‘step(a-Tstat) + 2*step(b-Tstat)’

Also, what’s the difference between running:

-gltsym ‘SYM: +Both -Left \ +Both -RIght’

or running them separately, i.e.

-gltsym 'SYM: +Both -Left
-gltsym 'SYM: +Both -Right

When I ran the former it gave me one Fstat and two different Coef and Tstat contrasts that were similar to the latter. When I run the later it will output two different Fstats.

Thanks again for all of your help.

6

For conjunction, the following should be enough:

3dcalc -prefix conjunction_map -a contrast1 -b contrast2 -expr ‘step(a-Tstat) *step(b-Tstat)’

The two t-statistics from

  1. -gltsym ‘SYM: +Both -Left \ +Both -RIght’

are the same as

  1. -gltsym 'SYM: +Both -Left
    -gltsym 'SYM: +Both -Right

The F-stat from 1) is associated with the hypothesis H_0: Both = Left and Both = Right, while the two F-statistics from 2) are the two t-statistics squared.