To date, surprisingly little attention has been directed toward determining
the optimum TR in a functional imaging experiment. A survey of the literat
ure reveals a wide range of TRs, but little justification for a specific TR
. Long-TR functional imaging experiments provide maximum signal-to-noise ra
tio (SNR) in the raw images; allow for the collection of a large number of
slice locations; and decrease the size of the data set acquired, simplifyin
g storage and handling. This work, however, demonstrates that long-TR imagi
ng sacrifices statistical power when the paradigm timing is held fixed. Tha
t is, for a fixed-run duration consisting of multiple activation/control bl
ocks, shorter TR acquisitions (on the order of 1000 ms) provide better disc
rimination between the activated and nonactivated brain tissue regions than
do long-TR acquisitions (on the order of 4000 ms). Results are shown for m
odeling the functional imaging experiment and for three different paradigms
performed on normal subjects. (C) 2001 Wiley-Liss, Inc.