Oh oh oh. Got it. I thought the threshold was applied to the input values. Not the output.
Things are fine when I set the threhold to something very small, eg, '-t -999'.
Thanks for the clarification and easy 'solution'.
Warm regards,
* ba
> Hi - with this option turned on the sign of the correlation is kept - so if the correlation is negative it won't pass the thresholding, and won't get reported.
> Cheers.
>
>> On 28 Sep 2012, at 06:39, bettyann wrote:
>>
>> Hello all,
>>
>> I see unexpected behavior when I use 'fslcc'. I get no output when I use the '--noabs' switch.
>>
>> I want to see if two eigenvariate time courses I extracted via 'fslmeants' are flipped with respect to one another. I extract the time courses:
>>
>> fslmeants -i filtered_func \
>> -m hippoLeft -o tc_eig_hippoLeft.txt --eig
>>
>> fslmeants -i filtered_func \
>> -m hippoRight -o tc_eig_hippoRight.txt --eig
>>
>> I convert these ascii files to .nii:
>>
>> nn=`fslnvols filtered_func `
>> fslascii2img tc_eig_hippoLeft.txt \
>> $nn 1 1 1 1 1 1 1 vol_eig_tc_eig_hippoLeft
>> fslascii2img tc_eig_hippoRight.txt \
>> $nn 1 1 1 1 1 1 1 vol_mean_tc_eig_hippoRight
>>
>> Then compute the cross correlation without and with the --noabs switch:
>>
>> # without --abs switch
>> fslcc -t 0 -p 6 vol_eig_tc_eig_hippoLeft vol_mean_tc_eig_hippoRight
>> 1 1 0.838791
>>
>> # with --abs switch
>> fslcc --noabs -t 0 -p 6 vol_eig_tc_eig_hippoLeft vol_mean_tc_eig_hippoRight
>>
>> When I plot these two eigenvariates, I can see they are anti-correlated. In the above example, when using the --noabs switch I expected a negative value, -0.838791. Instead, I get no output.
>>
>> Just in case, I also tried --noabs=1 and --noabs=0 and got no output. I also moved the position of the --noabs switch within the command -- still no output. Any suggestions?
>>
>> I am trying to avoid popping into matlab.
>>
>> Thanks,
>> * ba
>>
|