Skip to content

Latest commit

 

History

History
73 lines (57 loc) · 2.51 KB

Statistics.md

File metadata and controls

73 lines (57 loc) · 2.51 KB

for ## Statistics documentation


Calculate subject-level statistics

with Stats.sh

zsh Stats.sh sub-01 wnw


Extract the Beta coefficients & R-squared values from the REML stats file

3dTcat -prefix Stats/Coefficients stats.${sub}.${g}_REML+orig'[2..$(4)]'
3dTcat -prefix Stats/R2 stats.${sub}.${g}_REML+orig'[4..$(4)]'


Calculate the contrast-to-noise ratio: Beta coefficient / noise residual

3dcalc -a Coefficients+orig'[7]' -b ../noise.all+orig -expr 'a/b' -prefix CNR_WNW
3dcalc -a Coefficients+orig'[8]' -b ../noise.all+orig -expr 'a/b' -prefix CNR_VisAud


Transform R-squared values to standardized, zero-mean Fisher Z-scores

3dcalc -overwrite -a R2+orig'[7]' -b Coefficients+orig'[7]' -expr 'atanh(sqrt(a))*(ispositive(b)-isnegative(b))' -prefix FisherZ_WNW

3dcalc -overwrite -a R2+orig'[8]' -b Coefficients+orig'[8]' -expr 'atanh(sqrt(a))*(ispositive(b)-isnegative(b))' -prefix FisherZ_VisAud


Overlay ROI mask over each FisherZ file to calculate the average FisherZ scores

3dROIstats -nzvoxels -nobriklab -mask ${root}Proc_Anat/StudyROIs/${sub}.FuncROIs.nii.gz FisherZ_WNW+orig >> Avg_FisherZROIs_WNW.1D

3dROIstats -nzvoxels -nobriklab -mask ${root}Proc_Anat/StudyROIs/${sub}.FuncROIs.nii.gz FisherZ_VisAud+orig >> Avg_FisherZROIs_VisAud.1D


Scrape the degrees of freedom from the .txt file from Stats

DOF=`3dAttribute BRICK_STATAUX stats.${sub}.${g}_REML+orig'[0]' | awk '{print $5}'`
echo $DOF >> Stats/DOF.txt


Calculate the contrast-to-noise ratio for all ROIs and report number of voxels in each ROI

3dROIstats -nzvoxels -nobriklab -mask ${root}Proc_Anat/StudyROIs/${sub}.FuncROIs.nii.gz CNR_WNW+orig >> CNR_ROIs_WNW.1D

3dROIstats -nzvoxels -nobriklab -mask ${root}Proc_Anat/StudyROIs/${sub}.FuncROIs.nii.gz CNR_VisAud+orig >> CNR_ROIs_VisAud.1D


Calculate group-level statistics

with groupstats.py

python3 groupstats.py wnw


Outputs: Pandas Dataframe with Subjects (rows), ROIs (columns), GLM (file name)
This will create group pandas .csv files for each output statistic for creating the "Visualizations" graphs.

# Voxel counts
f"{out}{g}_Voxels_group.tsv"

# Fisher Z values
f"{out}{g}_FisherZ_WNW_group.tsv"
f"{out}{g}_FisherZ_VisAud_group.tsv"

# Degrees of Freedom
f"{out}DOF_GLMs_All_group.tsv"


To view graphical representations of the above results:

See ../../Visualizations/OHBM2022Poster/Visualizations.md