I was doing the analysis in MATLAB previously. I computed the t-statistic using BOLD values for Baseline condition Blocks Vs condition A Blocks for all runs across sessions. All the time series data in each run is expressed as percent of the mean (scaling) before the analysis.
Here is the 3dDeconvolve code I used for the regression: B0,FA0,PA0 are the baseline, FA, and PA blocks where monkey's fixation did not meet the criteria. FA PA are the regressors of interest. good baseline blocks are not modeled.
3dDeconvolve -input ${subj}_r???_scaled+orig.HEAD \
-polort 'A' \
-num_stimts 12 \
-stim_times 1 stim_times_rAll_c1v1.1D 'BLOCK(27.5,1)' \
-stim_label 1 FA \
-stim_times 2 stim_times_rAll_c2v1.1D 'BLOCK(27.5,1)' \
-stim_label 2 PA \
-stim_times 3 stim_times_rAll_c0v0.1D 'BLOCK(27.5,1)' \
-stim_label 3 B0 \
-stim_times 4 stim_times_rAll_c1v0.1D 'BLOCK(27.5,1)' \
-stim_label 4 FA0 \
-stim_times 5 stim_times_rAll_c2v0.1D 'BLOCK(27.5,1)' \
-stim_label 5 PA0 \
-stim_file 6 SC_R_rAll_ts.1D -stim_label 6 SC_RInt \
-stim_file 7 ${subj}_rAll_motion.1D'[1]' -stim_base 7 -stim_label 7 roll \
-stim_file 8 ${subj}_rAll_motion.1D'[2]' -stim_base 8 -stim_label 8 pitch \
-stim_file 9 ${subj}_rAll_motion.1D'[3]' -stim_base 9 -stim_label 9 yaw \
-stim_file 10 ${subj}_rAll_motion.1D'[4]' -stim_base 10 -stim_label 10 dS \
-stim_file 11 ${subj}_rAll_motion.1D'[5]' -stim_base 11 -stim_label 11 dL \
-stim_file 12 ${subj}_rAll_motion.1D'[6]' -stim_base 12 -stim_label 12 dP \
-gltsym 'SYM: PA -FA' \
-glt_label 1 PA-FA \
-fout -tout -x1D X.xmat.1D -xjpeg X.jpg \
-fitts fitts.$subj \
-errts errts.$subj \
-bucket stats.$subj
