Computes the DSE decomposition and DVARS-related statistics. Citation: Insight and inference for DVARS (Afyouni and Nichols, 2018)

Source: R/DVARS.R

Differences from implementation at github.com/asoroosh/DVARS:

  • The matrix is transposed.

  • We center and scale the matrix differently (see scale_med)

  • We set all zero-variance voxels to zero during centering & scaling. This means that when we remove constant 0 or NA voxels, constant non-zero voxels are also removed.

  • We use a tolerance of \(1e-8\) to detect non-zero voxels.

DVARS(X, normalize = FALSE, norm_I = 100, verbose = FALSE)

Arguments

X

a T x N numeric matrix representing an fMRI run.
normalize

Normalize the data as proposed in the original paper? Default is FALSE.
norm_I

The value to scale to. Default is 100, as in the original paper.
verbose

Should occasional updates be printed? Default is FALSE.