I wonder if it would be possible to add a couple of features:
1. In 3dttest, it is possible to test datasets against a set value. The assumption is that that value is a known quantity with zero error. Would it be possible to allow an associated variance to be specified? Sometimes the value that we're testing against is itself an estimate, so it would be more correct to use the associated variance when performing the t-test.
2. Would it be possible to include a similar option in 3dDeconvolve? i.e. when specifying GLT contrasts, it would be nice to test a given contrast or condition against a constant value (e.g. a percent signal change value taken from previous studies/different group/different part of the brain*). As with the t-test, this would preferably allow an associated variance to be specified as well.
Hope this makes sense. This feature comes from us trying to get away from simple binary activated versus non-activated tests, and at least make an effort to use more quantitative comparisons. With EPI there are of course problems, but with perfusion based FMRI these sorts of tests become more appropriate.
Thanks!
*difficulties of comparing activation from different parts of the brain acknowledged.