Literature review on fMRI outlier detection

[sdiebolt]

We can use this issue to suggest and discuss methods for fMRI outlier detection!

[FelipeCybis]

In Powell et al., 2012 we have good description of DVARS (which we have already implemented) and framewise displacement (FD) metrics, which would be good to test.

FD metric minimizes the registration error of $I_i$ using the rigid body transforms $T_i$ against a reference frame $I_0$:

$$ \begin{align} \varepsilon_i = \text{mean}( (sI_i(T(\vec{x}) - I_0(\vec{x}))^2) \end{align} $$

Where $T_i(\vec{x})$ is the combination of the rotation and displacement matrices in the 3 dimensions:

$$ \begin{align} T_i = \left\lbrack \matrix{R_i & d_i \cr 0 & 1} \right\rbrack \end{align} $$