rcTrainTranslation: rcTrainTranslation

In stnava/rcissus: Fast Segmentation Or Image Translation With Deep Learning

Description

Training for rcissus translation between image modalities

Usage

rcTrainTranslation(x, y, masks, nBases = 6, patchRadius = 3,
  meanCenter = FALSE, nsamples = 1000, seeds, hidden = c(200, 200),
  classification = FALSE, dataScaling = 1000, batchSize,
  mdlMethod = "h2o", epochs = 10)

Arguments

x	input filenames or image list defining source modality
y	input filenames or image list defining target modality
masks	input filenames or image list defining source modality masks
nBases	number of eigenvector bases
patchRadius	patch radius, integer value
meanCenter	boolean
nsamples	number of samples per image
seeds	random seeds for reproducibility across testing modalities
hidden	Hidden layer sizes (e.g. c(100,100)). Defaults to c(200,200).
classification	boolean
dataScaling	scaling by which to divide input data - results may be sensitive to this value
batchSize	for training
mdlMethod	string either lm, h2o or kerasdnn
epochs	number of epochs (integer) over which to train

Value

list containing the image bases and the trained model and other relevant parameters.

Author(s)

Avants BB

stnava/rcissus documentation built on May 12, 2019, 6:25 p.m.