3dLME GLT coding

AFNI Message Board
1

Hi,

I would like to ask about 3dLME GLT coding.
I just bring the ‘example 3’ from ‘3dLME’ instruction on the web as following.

3dLME -prefix Example3 -jobs 24
-model "condgroup"
-ranEff ‘~1’
-SS_type 3
-num_glt 6
-gltLabel 1 ‘pos-neu’ -gltCode 1 'cond : 1pos -1neu’
-gltLabel 2 ‘neg’ -gltCode 2 'cond : 1neg ’
-gltLabel 3 ‘pos+nue-neg’ -gltCode 3 ‘cond : 1pos +1neu -1neg’
-gltLabel 4 ‘pat_pos-neu’ -gltCode 4 'cond : 1pos -1neu group : 1pat’
-gltLabel 5 ‘pat_neg-neu’ -gltCode 5 ‘cond : 1neg -1neu group : 1pat’
-gltLabel 6 ‘pat_pos-neg’ -gltCode 6 'cond : 1pos -1neg group : 1pat’
-num_glf 1
-glfLabel 1 ‘pos-neu’ -glfCode 1 ‘Group : 1ctr & 1pat cond : 1pos -1neu & 1pos -1neg’
-dataTable
Subj cond group InputFile
s1 pos ctr s1+tlrc’[pos#0_Coef]’
s1 neg ctr s1+tlrc’[neg#0_Coef]’
s1 neu ctr s1+tlrc’[neu#0_Coef]’
…
s21 pos pat s21+tlrc’[pos#0_Coef]’
s21 neg pat s21+tlrc’[neg#0_Coef]’
s21 neu pat s21+tlrc’[neu#0_Coef]’ \

My question is if these following GLT coding would provide the identical t-test result.

  1. -gltLabel 1 ‘pos-neu’ -gltCode 1 ‘cond : 1pos -1neu’ \

  2. -gltLabel 7 ‘pos+nue-neg+neg’ -gltCode 7 'cond : 1pos +1neu -1neg -1neu ’ \

I assumed that in the second case condition ‘neu’ should be cancelled out.
I have tried this way with my data, but these two gave me the different result.
Could you let me know what happen in this coding?

Thanks

Jung

2

Are they slightly or dramatically different?

3

It gave totally different result.

Jung

4

Jung,

The coding

  1. -gltLabel 7 ‘pos+nue-neg+neg’ -gltCode 7 'cond : 1pos +1neu -1neg -1neu ’ \

is not interpreted correctly by the program. So, use the following

  1. -gltLabel 1 ‘pos-neu’ -gltCode 1 ‘cond : 1pos -1neu’ \
5

Thanks.

In this case, if I use contrast image, for example ’ pos_neu = pos -neu’ and ’ nag_neu= neg- neu’
and apply them to 3dLME,

a) -gltLabel 1 ‘pos-neu’ -gltCode 1 ‘cond : 1pos -1neu’ \

b) -gltLabel 8 ‘pos-neg_contrast’ -gltCode 8 ‘cond : 1pos_neu -1neg_neu’ \

I expected the similar result, but it also gave totally different result.

Can I get some idea about this?

Jung

6

f I use contrast image, for example ' pos_neu = pos -neu' and ' nag_neu= neg- neu' and apply them to 3dLME

It's hard for me to understand how exactly you're implementing this in 3dLME. Could you provide the 3dLME scripts to show the details?