3dTcorrMap output for “positive correlation” or “negative correlation”

AFNI Message Board
1

Hello，

I want to use 3dTcorrMap to calcualte the global connectivity.

Given that mixed positive and negative value could cancel each other( eg average 0.1 and -0…1). so I want to average positive correaltion and negative correlation separately.

3dTcorrMap -input input_data -mask mask_data -Cexpr ‘step(z-0)*z’ output_data ( for Zmean)
3dTcorrMap -input input_data -mask mask_data -Cexpr ‘step(r-0)*r’ output_data (for rmean)

Is this right for postive correlation?

how to analysis the negative correlation?

Thank you so much!

Best,

jinsong

2

Jinsong,

For positive correlation, try

3dTcorrMap -input input_data -mask mask_data -Cexpr ‘step(r)*r’ output_data

Similarly for negative correlatoin:

3dTcorrMap -input input_data -mask mask_data -Cexpr ‘step(-r)*r’ output_data

3

Hi Gang,

Thank you so much!

I tried both r and z :
3dTcorrMap -input input_data -mask mask_data -Cexpr ‘step(r)*r’ r_output_data
3dTcorrMap -input input_data -mask mask_data -Cexpr ‘step(z)*z’ z_output_data

But, the values in the z_output_data seems a slightly higher than that in the r_output_data. So, what is the difference between these two?

Thanks,

Jinsong

4

the values in the z_output_data seems a slightly higher than that in the r_output_data. So, what is the difference between these two?

My guess is that the z_output_data is averaged after the Fisher transformation while the r_output_data is averaged without any transformation.

5

Hi Gang,
Thank you for your help!