Calculate Moments of Neighborhood

This function calculates Local Moments (mean, standard deviation, skew) for an array.

local_moment(
  image,
  window = NULL,
  nvoxels = NULL,
  moment,
  mask = NULL,
  only.mask = FALSE,
  center = is.null(mean_image),
  invert = FALSE,
  mean_image = NULL,
  na.rm = TRUE,
  remask = TRUE,
  ...
)

Arguments

image	input image
window	window (in width) for the neighborhood
nvoxels	window (in voxels) for the neighborhood 1 results in a 3x3 cube
moment	vector moments taken (1- mean, 2-sd, 3-skew)
mask	array or object of class nifti of same size as image
only.mask	Should objects outside the mask (i.e. zeros) be counted the moment? Default is FALSE so edges are weighted to 0
center	vector of indicator of central moment. if TRUE mean image is subtracted. Same length as moment
invert	Standardize the values by the power: 1/moment
mean_image	mean image to be subtracted. If not supplied, and central = TRUE, local_moment_edge is run with mom = 1
na.rm	remove NAs from the moment image calculation
remask	set areas outside of mask to 0
...	Arguments passed to get_neighbors

Value

List of arrays same lenght as moment

Examples

x = array(rnorm(1000), dim=c(10, 10, 10))
mask = abs(x) < 1
mean.x = local_moment(x, nvoxels=1, moment = 1, mask=mask,
center = FALSE,
remask = FALSE)[[1]]
var.x = local_moment(x, nvoxels=1, moment = 2, mask=mask, center = TRUE,
mean_image = mean.x, remask=FALSE)[[1]]

### check that x[2,2,2] mean is correct
check = x[1:3,1:3,1:3]
## masking
vals = check[abs(check) < 1]
m = mean(vals)
all.equal(m, mean.x[2,2,2])
#> [1] "Mean relative difference: 0.4074074"
n = length(vals)
v = var(vals) * (n-1)/n
var.x[2,2,2]
#> [1] 0.2301687
all.equal(v, var.x[2,2,2])
#> [1] "Mean relative difference: 0.4073575"