Next: Constraining Basis Function Linear
Up: Model
Previous: Autoregressive Parameters Spatial Prior
In this work we model the different HRF shapes for different
underlying conditions at different voxels using basis functions
and assuming a linear time invariant system (Josephs et al., 1997). If
we have
underlying conditions, and for each
condition we have
basis functions to model the HRF
for that condition for voxel we can rewrite
equation 1 as:
where
is the regression parameter for the
basis function of the underlying condition at voxel
and:
where represents convolution, is the
basis function and is the stimulus
function. Note that is an index at higher temporal
resolution than , to capture all of the HRF shape (i.e. for a
resolution of of a TR,
). We obtain
from
by sampling every
.
Next: Constraining Basis Function Linear
Up: Model
Previous: Autoregressive Parameters Spatial Prior