PowerPoint Presentation - Serial Correlation in AFNI

–13–

Tentative Conclusions

• For individual subject thresholded activation maps:

H Use GLSQ/REML estimation, especially for slow block design experiments at 3+ Tesla

H Be aware that there may be many false negatives

o i.e., false acceptances of the null hypothesis

o am looking into an FDR-like procedure for estimating the false negative rate, similar to how FDR estimates the false positive rate

• For group maps using ANOVA (or similar statistics):

H Differences between OLSQ and GLSQ estimation are small

• Recommendations:

H Don’t need to re-visit group activation conclusions!

H Use 3dREMLfit as a near drop-in replacment for 3dDeconvolve for future work

oA little extra CPU time (usually from 1..3 times as long)