ROI-Based Transformation of Neuroimages into High-Dimensional Data Frames.

MRIreduce

Introduction

MRIreduce is an advanced R package designed to convert NIfTI format T1/FL neuroimages into structured, high-dimensional 2D data frames, focusing on region of interest (ROI) based processing. This package incorporates a key algorithm called ‘partition’, which offers a fast, flexible framework for agglomerative partitioning based on the Direct-Measure-Reduce approach. This method ensures that each reduced variable maintains a user-specified minimum level of information while being interpretable, as each maps uniquely to one variable in the reduced dataset. The ‘partition’ algorithm, detailed in Millstein et al. (2020), allows for customization in variable selection, measurement of information loss, and data reduction methods. MRIreduce is indispensable for researchers requiring efficient, accurate preparation of neuroimaging data for detailed statistical analysis and machine learning applications, enhancing the interpretability and utility of neuroimaging studies.

Installation Instructions

This document provides detailed steps to install the necessary dependencies for the package. Please follow the instructions carefully to ensure all dependencies are correctly installed.

Step 1: Install devtools

The devtools package is essential for installing packages directly from GitHub. If you do not have devtools installed, run the following code:

if (!requireNamespace("devtools", quietly = TRUE))
    install.packages("devtools")

Step 2: Install fslr

The fslr package is an interface to the FSL (FMRIB Software Library) tools. Install it from CRAN:

install.packages("fslr")

Step 3: Optionally install EveTemplate

If you want the package to resolve the EVE template automatically, install the optional EveTemplate package. Otherwise, you can supply a template path directly to eve_T1(), eve_Fl(), and map2_eve().

Install EveTemplate from the Neuroconductor repository:

install.packages(
  "EveTemplate",
  repos = c(
    "https://neuroconductor.org/releases/2020/05",
    getOption("repos")
  )
)

Step 4: Install FSL

FSL is not an R package but a standalone software suite for MRI and fMRI analysis. Follow the instructions on [their website] (https://fsl.fmrib.ox.ac.uk/fsl/docs/#/install/index) to download and install it.

Step 5: Optional Python setup for map2_eve

The map2_eve() helper also requires a working Python environment for reticulate with the Python packages nilearn, nibabel, and matplotlib. This is only needed if you want to generate EVE-overlay figures from R.

Notes on external requirements

• The package can be installed and its core pure-R functionality can be used without FSL or Python.
• eve_T1() and eve_Fl() require FSL.
• map2_eve() requires Python packages configured for reticulate.
• If EveTemplate is not installed, you can still use the relevant functions by supplying template_img_path directly.