Skip to contents

Compute a brain-like mask from image intensities

Source: R/all_generic.R, R/filebacked_neurovec.R, R/mapped_neurovec.R, and 2 more
automask-methods.Rd

Builds a spatial mask from image content rather than requiring an external mask. The implementation is AFNI-inspired: it computes a clip level, applies thresholding to a representative 3D image, retains the largest connected component, and optionally applies peel/unpeel cleanup.

Usage

automask(
  x,
  mfrac = 0.5,
  gradual = TRUE,
  representative = "mean_abs",
  peels = 1L,
  peel_threshold = 17L,
  connect = c("26-connect", "18-connect", "6-connect")
)

# S4 method for class 'FileBackedNeuroVec'
automask(
  x,
  mfrac = 0.5,
  gradual = TRUE,
  representative = "mean_abs",
  peels = 1L,
  peel_threshold = 17L,
  connect = c("26-connect", "18-connect", "6-connect")
)

# S4 method for class 'MappedNeuroVec'
automask(
  x,
  mfrac = 0.5,
  gradual = TRUE,
  representative = "mean_abs",
  peels = 1L,
  peel_threshold = 17L,
  connect = c("26-connect", "18-connect", "6-connect")
)

# S4 method for class 'DenseNeuroVec'
automask(
  x,
  mfrac = 0.5,
  gradual = TRUE,
  representative = "mean_abs",
  peels = 1L,
  peel_threshold = 17L,
  connect = c("26-connect", "18-connect", "6-connect")
)

# S4 method for class 'AbstractSparseNeuroVec'
automask(
  x,
  mfrac = 0.5,
  gradual = TRUE,
  representative = "mean_abs",
  peels = 1L,
  peel_threshold = 17L,
  connect = c("26-connect", "18-connect", "6-connect")
)

# S4 method for class 'NeuroVol'
automask(
  x,
  mfrac = 0.5,
  gradual = TRUE,
  representative = "mean_abs",
  peels = 1L,
  peel_threshold = 17L,
  connect = c("26-connect", "18-connect", "6-connect")
)

Arguments

x

A neuroimaging object.

mfrac

Fraction used in the clip-level update step.

gradual

If TRUE, use a spatially varying clip map across octants. If FALSE, use a single global clip threshold.

representative

For multi-volume inputs, the 3D summary image used for thresholding. Supported values are "mean_abs" and "median".

peels

Number of AFNI-style peel/unpeel iterations. Use 0 to disable.

peel_threshold

Minimum number of 18-neighbors required for a voxel to survive peeling.

connect

Connectivity used when retaining the largest component.

Value

A LogicalNeuroVol.

Examples

sp <- NeuroSpace(c(8, 8, 8), spacing = c(1, 1, 1))
arr <- array(abs(rnorm(512)), c(8, 8, 8))
arr[3:6, 3:6, 3:6] <- arr[3:6, 3:6, 3:6] + 10
vol <- NeuroVol(arr, sp)
mask <- automask(vol, gradual = FALSE, peels = 0)