Brief look at header information of stats data
useful information like string labels associate with each sub-brick
degrees of freedom and other useful information saved
save both coefficient ('beta' or effect estimate) and associated (T-)stat
Overlay the Coef information
use AFNI-recommended 'scale' during processing to interpret Coef
Coef has interpretable units of local BOLD % signal change
we strongly recommend displaying the Coef values
make meaningful cbar ranges
probably ~3% for block or strong tasks
maybe ~1% for event or subtle tasks
is it bad to see large Coef outside brain (???)
no, because FMRI is veeery noisy
we use stat information to focus on high significance (low std err) voxels
we also often use group information to reduce noise effects
Threshold using the Tstat information
Tstat = (effect estimate) / (standard err)
with DF info, can convert Tstat to p-value (significance)
since we store DF info in AFNI header, can do this calc in GUI
so can threshold voxelwise stat as a p-value
Important to present ALL data from modeling
use threshold (of Tstat) to choose where to see stat info
use overlay (of Coef, not Tstat again!) to see effect estimate of model
Can see sub-threshold info still, usefully
thresholding helps to highlight strong results
but we don't want to lose info or hide artifacts, if we can help it
click the 'A' button above colorbar to see sub-threshold info
'alpha' channel is turned ON (and can be toggled)
suprathreshold = opaque; subthreshold = increasingly transparent
click hte 'B' button above colorbar to put box around suprathreshold values
helps distinguish sub/suprathreshold values more clearly