Generate segmentation-based augmentation data with on disk storage

generateDiskData(
  inputImageList,
  segmentationImageList,
  segmentationNumbers,
  selector,
  addCoordConv = TRUE,
  segmentationsArePoints = FALSE,
  maskIndex,
  smoothHeatMaps = 0,
  numpynames,
  numberOfSimulations = 16,
  referenceImage = NULL,
  transformType = "rigid",
  noiseModel = "additivegaussian",
  noiseParameters = c(0, 0.002),
  sdSimulatedBiasField = 5e-04,
  sdHistogramWarping = 5e-04,
  sdAffine = 0.2
)

Arguments

inputImageList	list of lists of input images to warp. The internal list sets contains one or more images (per subject) which are assumed to be mutually aligned. The outer list contains multiple subject lists which are randomly sampled to produce output image list.
segmentationImageList	of segmentation images corresponding to the input image list.
segmentationNumbers	the integer list of values in the segmentation to model
selector	subsets the inputImageList and segmentationImageList (eg to define train test splits)
addCoordConv	boolean - generates another array with CoordConv data
segmentationsArePoints	boolean - converts segmentations to points
maskIndex	the entry within the list of lists that contains a mask
smoothHeatMaps	numeric greater than zero will cause method to return heatmaps. the value passed here also sets the smoothing parameter passed to `smoothImage` in pixel/voxel space.
numpynames	the names of the numpy on disk files should contain something with the string mask if using maskIndex and something with the word coordconv if using CC. should include something with the word heatmaps if using heatmaps.
numberOfSimulations	number of output images. Default = 10.
referenceImage	defines the spatial domain for all output images. If the input images do not match the spatial domain of the reference image, we internally resample the target to the reference image. This could have unexpected consequences. Resampling to the reference domain is performed by testing using `antsImagePhysicalSpaceConsistency` then calling `resampleImageToTarget` upon failure.
transformType	one of the following options `c( "translation", "rigid", "scaleShear", "affine"," deformation" , "affineAndDeformation" )`.
noiseModel	one of the following options `c( "additivegaussian", "saltandpepper", "shot", "speckle" )`
noiseParameters	'additivegaussian': `c( mean, standardDeviation )`, 'saltandpepper': `c( probability, saltValue, pepperValue)` , 'shot': scale, 'speckle': standardDeviation. Note that the standard deviation, scale, and probability values are max values and are randomly selected in the range `[0, noise_parameter]`. Also, the "mean", "saltValue" and pepperValue" are assumed to be in the intensity normalized range of `[0, 1]`.
sdSimulatedBiasField	Characterize the standard deviation of the amplitude.
sdHistogramWarping	Determines the strength of the bias field.
sdAffine	Determines the amount of transformation based change

Value

list of array

Author

Avants BB

Examples

library( ANTsR )
#> Loading required package: ANTsRCore
#> ANTsRCore 0.7.5
#> For reproducible results, please set the environment variable ANTS_RANDOM_SEED,
#> either in .Renviron or with a seed (e.g. XXX): 
#> Sys.setenv(ANTS_RANDOM_SEED = XXX)
#> Sys.setenv(ITK_GLOBAL_DEFAULT_NUMBER_OF_THREADS = 1)
#> may also be required for reproducible results. See
#> https://github.com/ANTsX/ANTs/wiki/antsRegistration-reproducibility-issues
#> for more information
#> 
#> Attaching package: ‘ANTsRCore’
#> The following objects are masked from ‘package:stats’:
#> 
#>     sd, var
#> The following objects are masked from ‘package:base’:
#> 
#>     all, any, apply, max, min, prod, range, sum
ilist = list( list( ri(1) ), list( ri(2) ) )
slist = list(
  thresholdImage( ilist[[1]][[1]], "Otsu",3),
  thresholdImage( ilist[[2]][[1]], "Otsu",3) )
npn = paste0(tempfile(), c('i.npy','s.npy','heatmap.npy','coordconv.npy') )
temp = generateDiskData( ilist, slist, c(0:3), c(TRUE,TRUE), numpynames = npn  )
temp = generateDiskData( ilist, slist, c(0:3), c(TRUE,TRUE),
  segmentationsArePoints=TRUE, numpynames = npn )
temp = generateDiskData( ilist, slist, c(0:3), c(TRUE,TRUE),
  segmentationsArePoints=TRUE, smoothHeatMaps = 3.0, numpynames = npn )