Time averaged T2*
tat2 (code) wraps around 3dROIstats, 3dcalc, and 3dTstat to reduce 4D EPI BOLD data to a per-voxel (3D) measure (nT2*) that is inversely related to iron concentration.
This is used in
- Contributions of dopamine-related basal ganglia neurophysiology to the developmental effects of incentives on inhibitory control
- In vivo evidence of neurophysiological maturation of the human adolescent striatum
And is initially from Predicting Individuals' Learning Success from Patterns of Pre-Learning MRI Activity
The normalization is within the brain for each time point. Not all voxels in the volume, but only those that belong to the brain mask. Then we normalize so that the sum of all brain voxels is some fixed number, e.g., 10000. The number doesn't really matter.
Also see Relative Concentration of Brain Iron (rcFe).
Setup
See the setup instructions for all of lncdtools. Breifely, git clone https://github.com/lncd/lncdtools and add the new directory to your path.
The raw tat2 script can stand alone, but will uses other lncdtools scripts if avaiable -- namely gitver. It's also a lot easier to fetch and track updates when the script is within source control. You get that when you clone the repo.
Usage
see tat2 --help
Simple
With options and relative paths
sub_ses="sub-01/ses-01"
censor_regex='s/.*func\/(.*)-preproc_bold.nii.gz/censor_files\/\1\/preproc_censor-fd0.3.1D/'
tat2 \
-output "deriv/$sub_ses/func/${sub_ses//\//_}_space-MNI152NLin2009cAsym_tat2.nii.gz" \
-mask_rel 's/preproc_bold.nii.gz/brain_mask.nii.gz/' \
-censor_rel "$censor_regex" \
-median_time \
-median_vol \
-no_voxscale \
-verbose \
deriv/$sub_ses/func/*space-MNI152NLin2009cAsym_desc-preproc_bold.nii.gz
Preprocessing
We slice-time and motion correction, skull strip, despiking (wavelet), and warp to MNI before running tat2.
Notably, smoothing is not included in datasets input to tat2.
Pipeline
Example
Example mean tat2 images 1
Comparisons
Permutation of tat2 calls were compared against R2 acquisitions.
-vol_median is likely the appropriate normalization.
Correlation with R2
tat2 is negatively correlated R2* 2
