This document describes the use of the dl4ni
package, Deep Learning Flows for Neuroimaging.
Recent years have seen the rise of Deep Learning applied to Neuroimgining problems. Some concepts and state-of-the-art are reviewed here and here.
There are a lot of packages in the R ecosystem focused on several aspects of Neuroimage processing, many of them listed in the CRAN Task View: Medical Imaging section: oro.nifti
, oro.dicom
, RNifti
, fmri
, divest
(among others) for reading/writing of images; ANTsR
, fslr
for more advanced processing methods; RNiftyReg
specialized in registration, mritc
focused in image segmentation. Many more, for DTI, fMRI, molecular imaging, and visualization are available in CRAN.
For Deep Learning, there are basically 4 packages maintained by diverse organizations: mxnet
from Apache MXNet, h2o
from H2O.ai and tensorflow
and keras
from Google and RStudio. All of these packages are available at CRAN.
The lack of packages to apply Deep Learning techniques to Neuroimaging pipelines inspired the creation of the dl4ni
package.
This package has been written and is developed with some key features in mind:
R
functions (including those from other packages, such as ANTsR
or fslr
) and Deep Learning models. This package is based in the keras
package, so a working installation of both tensorflow
and keras
is a minimum requirement. Check the package website for detailed instructions on how to get those packages installed and running.
Currently, this package is not in CRAN. The only way to get this package installed is by obtaining the source and building and installing the package in a local system.
#!bash
R CMD INSTALL /path/to/dl4ni
Feel free to use any of the provided demos in the package. Just load the package and list the available demos as:
#!R
library(dl4ni)
list.files(system.file("demo", package = "dl4ni"))
And open the one you like most!
To know more about the use of this package, head to the wiki.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.