predict_flexconn: Predict from FLEXCONN model
In muschellij2/flexconnr: 'FLEXCONN' Model Wrapped in R

Description Usage Arguments Value Examples

View source: R/predict_flexconn.R

Predict from FLEXCONN model

1
2
3

predict_flexconn(t1, flair, t2 = NULL, outdir = NULL, gpu = "cpu",
  normalize = TRUE, num_atlases = c("21", "61"),
  outcomes = c("mask1", "mask2"), verbose = TRUE)

`t1`	T1-weighted image to predict from: skullstripped, bias-corrected). Since the training is 2D, make sure the test image is properly oriented, i.e. the in-plane has the highest native resolution. E.g. the training images are axial because their native resolution is 1x1x4mm^3 in axial RAI orientation.
`flair`	FLAIR image to predict from, must be registered to T1 and have same orientation as T1
`t2`	(optional) T2 image to predict from, must be registered to T1 and have same orientation as T1
`outdir`	Output directory for predictions
`gpu`	Either an integer for the GPU. Use "cpu" to use CPU.
`normalize`	Should the images be normalized?
`num_atlases`	Specifies which model to use. Determined by the number of atlases in the FLEXCONN model.
`outcomes`	The outcome used to train the model, from rater 1 or rater 2
`verbose`	Print diagnostic messages

A vector of filenames

# predict_flexconn(python_cmd = "python3)
library(reticulate)
## Not run: 
reticulate::use_python("/Library/Frameworks/Python.framework/Versions/3.5/bin/python3")
# reticulate::use_python("python3")

flair = system.file("extdata", "FLAIR.nii.gz", package = "flexconnr")
t1 = system.file("extdata", "T1.nii.gz", package = "flexconnr")
pp = predict_flexconn(t1 = t1, flair = flair)
# result = RNifti::readNifti(pp[2])

## End(Not run)