Classical and Bayesian Inference in fMRI - Advance Image Processing in Magnetic Resonance Imaging - page 543

contrast

100

SPM (T)

200

300

Height threshold T = 4.86
Extent threshold k = 0 voxels

12
Design matrix

34

z = 3 mm

FIGURE 17.5 Same as for Figure 17.4 , but this time showing the corresponding SPM

using a corrected threshold at p = 0.05.

(circled cluster on the PPM in the lower panel of Figure 17.4). The SPM is more
conservative because the correction for multiple comparisons in these data is very
severe, rendering classical inference relatively insensitive. It is interesting to note
that dynamic motion in the visual field has such widespread (though small) effects
at a hemodynamic level.

17.6 DYNAMIC CAUSAL MODELING

Dynamic causal modeling (DCM) (31) is used to make inferences about func-
tional integration from fMRI time series. The term “causal” in DCM arises
because the brain is treated as a deterministic dynamical system in which external
inputs cause changes in neuronal activity, which in turn cause changes in the
resulting BOLD signal that is measured with fMRI. This is to be contrasted with
a conventional GLM where there is no explicit representation of neuronal activity.
The second main difference to the GLM is that DCM allows for interactions
between regions. Of course, it is this interaction which is central to the study of
functional integration.