De-normalize 3dTproject output?

I’m currently denoising some functional images and looking to use 3dTproject, which has been easy enough. However, the output is normalized and I need to use the output to generate average functional activation for ROIs in a parcellation, which isn’t possible with the normalized output (i.e. average = 0). Is there a way to skip the normalization step at the end?


Would you provide more detail on what you want?

For example, if the problem is that the mean is zero, it would be simple to pre-compute the mean (3dTstat -mean) and add it back in later (3dcalc -a TIME_SERIES -b mean -expr a+b). However, that should not have any effect on computations of functional activation.

So exactly how do you plan to compute the functional activation for the ROIs after 3dTproject? Neither correlations nor regressions should be affected by the zero-mean data.

  • rick

Hi Rick,

Thank you for your response. I’m just trying to calculate the mean BOLD signal for a given ROI using fslstats after the data has been denoised. Right now I’m denoising with 3dBandpass without the norm flag but from the documentation I’ve read 3dTproject is strongly suggested over 3dBandpass.