Author: Sachi Patel
In the last few years, Resting-state Functional Magnetic Resonance Imaging (RS-fMRI or R-fMRI) has been drawing more attention. By measuring spontaneous low-frequency fluctuations in the blood-oxygen-level-dependent (BOLD) signal, RS-fMRI allows researchers to investigate the functional architecture of the brain. However, a publicly available, systemically integrated, and easy-to-use tool for RS-fMRI data processing was still lacking to researchers developed a toolkit in MATLAB for the analysis of RS-fMRI data: Resting-State fMRI data analysis Toolkit (REST).
REST is a user-friendly toolkit with a wide range of functions:
It is primarily used to perform statistical analysis and calculate:
Functional Connectivity (FC)
Regional Homogeneity (ReHo)
Amplitude of Low-Frequency Fluctuation (ALFF)
Fractional ALFF (fALFF)
Pearson linear correlation is also a widely used method in RS-fMRI analysis, it was also implemented in REST. Some additional functions were implemented in REST, including:
Linear trend removal
Regression of covariates
Time course extraction
Installing REST is relatively simple: download the most recent version from http://www.restfmri.net, extract the compressed package to a predefined directory, and then add the full path to MATLAB’s search path. REST is compatible with MATLAB version 6.5 or higher.
Almost all REST functions only need MATLAB's basic run time environment. Entering ‘‘rest’’ in the MATLAB command window will open REST’s GUI. The purple buttons are seen in the GUI window allowing access to the methods available for RS-fMRI data analysis -- important information about the main algorithms is all included below. The ‘‘Help’’ button in REST will also guide users to the RS-fMRI online forum (http://forum.restfmri.net) for more detailed information.
Main Algorithms in Rest
A. Functional Connectivity
Resting-state functional connectivity measures the “temporal correlation of spontaneous BOLD signals among spatially distributed brain regions,” assuming that regions that have correlated activity form functional networks. REST specifically uses Pearson linear correlation to calculate functional connectivity and effectively allows researchers to analyze RS-fMRI data.
The following image shows a demonstration of functional connectivity in REST
To compute functional connectivity, the user has to define an ROI and optional covariates (in this image the usage of covariates is not shown).
A seed ROI can be set by clicking the ‘‘Voxel wise’’ button seen in the image above. This allows the user to define parameters for calculating the linear correlation between a reference time course and the time course of either (1) each voxel in the brain or (2) voxels within a predefined mask. The “ROI wise” button, on the other hand, allows the user to define parameters for calculating the linear correlation between several ROIs and then calculate the correlation.
In this image specifically, two datasets were added. After that, the linear trend removal and band-pass (0.01 Hz,0.08 Hz) filtering options were then checked. A seed ROI was set by clicking the ‘‘Voxel-wise’’ button and the ‘‘Default mask’’ selection meant that REST automatically selected the default mask. The results of this computation allow researchers to come to conclusions regarding the functional interactions between two or more brain regions.
B. Regional Homogeneity (ReHo)
Resting-state activity is also often examined using regional homogeneity (ReHo). ReHo "estimates regional [brain] activation patterns through indices of a localized concordance" and essentially investigates the synchronization of spontaneous fMRI signals.
The following image shows a demonstration of regional homogeneity computation in REST:
The image is a display of the ReHO computation graphical user interface of REST. Specifically, computation of ReHo in REST done using Kendall’s coefficient of concordance (KCC). An individual ReHo map is obtained by calculating time course of voxel with those of its nearest neighbors.
REST includes a normalization process in the ReHo calculation that is similar to PET studies: an individual image is usually divided by the global mean value to minimize inter-individual variability. For PET studies, an individual image is usually divided by the global mean value to minimize inter-individual variability. Thus, the ReHo map could be divided by the mean ReHo value of voxels within a specific mask (whole brain mask or user-defined mask), which is the box that is checked at the bottom of the image.
The user can also select the cluster size and the mask. In this case, the default mask is selected and there is no linear trend removal and band-pass options checked. This should allow for the ReHo computation to be performed successfully.
C. Amplitude of Low Frequency Fluctuations (ALFF) & Fractional ALFF (fALFF)
ALFF is an RS-fMRI indicator that is used to detect the regional intensity of spontaneous fluctuations in the BOLD signal, which helps "identify the spontaneous neural activity of specific regions and physiological states of the brain".
The following image shows a demonstration of ALFF computation in REST:
After an individual ALFF map is obtained, a standardization/normalization procedure can be performed as done in the ReHo analysis (the individual ALFF map is divided by the mean ALFF value of voxels within a specific mask).
ALFF reflects the intensity of regional spontaneous brain activity. However, it is sensitive to the physiological noise found in some brain regions, such as in the cistern areas. Fractional ALFF (fALFF) is an improved ALFF approach that is implemented in REST.
In this image, there is no band pass filter set or linear trend removal, but when during ALFF analysis within rest, usually a specific frequency band is selected to generate an individual ALFF map. ALFF is defined as the "mean square root of the power spectrum density over the low-frequency band, which is usually 0.01~ 0.08 Hz. However, users can also choose specific frequency bands for their analysis (ex. a study found that the frequency band of 0.027 ~ 0.073 Hz is more specific to the basal ganglia).
Other Features in REST
D. REST DICOM Sorter
Data is produced by most MRI scanners in Digital Imaging and Communications in Medicine (DICOM) format. This feature in REST reads DICOM files by using existing routines (dicominfo and dicomread) and then classifies the DICOM files participant by participant and session by session. Within the DICOM files, the participants' private information (for example, their name, ID, and birthday) can be anonymized.
E. REST Slice Viewer
REST Slice Viewer is a routine for displaying results that can be overlaid on a structural image. There is a color bar that is designed to show the threshold and through the slice viewer, the user can see information about the underlay image and the overlay image including items such as dimension, voxel size, and origin. It includes common functions for image-processing and image displaying. It also reports information about different brain clusters including the number of voxels, anatomical term of the location, and peak intensity.
The Slice Viewer GUI is shown here:
F. Extract ROI Time Course
REST supports a routine to extract the mean time course of the voxels within a specific ROI. It could also be used to extract the time course of a single voxel. These time courses can be written as text files and the entire function is important in efficient ROI analysis.
G. Regress out covariates
REST includes a specific routine for regressing out some unnecessary covariants. The format for inputting covariants has to be in a text file with one or more columns that each represent a covariate. Many of the images of the computational graphical interfaces above do not involve this process, but it is an extremely useful feature of REST.
H. Power Spectrum
This is a utility that presents a voxel's time course and the power spectrum. The feature can be used to check the quality of the fMRI signal and there is an option included in REST to remove any linear trend in the time course before the power spectrum is calculated. Users just need to specify the folder that contains the functional images.