Title: Multiple comparisons in functional magnetic resonance imaging (fMRI)
In fMRI experiments,
neuroscientists are able to collect huge amounts of
data as a
given subject performs some cognitive task in the form of a
time series
of moderate-to-high resolution 3-dimensional images. This
time series
is often input into a regression model fit on a voxel-by-voxel
basis across
time. This results in an image of highly structured t or F statistics,
which presents
the data analyst with a large multiple comparisons problem.
In this talk,
we will compare a few approaches to this multiple
comparisons
problem, one using some results about the maximum of smooth
Gaussian random
fields, the other using the False Discovery Rate.