fMRI Chronic Pain Meta-analysis Documentation

Written by Anna Xu

Replicator: Alina Henn

Last Updated: May 12, 2020

Setup

For these analyses, you will need the following software:

To begin, download this GitHub repo. Next, move the spm12 and BrainNet Viewer folders into the dependencies folder. After you’re done with these steps, open the script runALE.m on MATLAB. This will be the script to run analyses. In the ale folder, create two subdirectories: DataMatlab and DataRaw. This is where inputted data will be saved later on.

The first line, curPath=genpath(pwd); addpath(curPath); will add all dependencies into your main directory. The next line, cd ale will put you in the ale folder so that you can begin your analyses.

For Windows OS

Please note that these scripts were ran using Mac OS (Mojave 10.14.1). Prior to running these scripts, if you are using Windows OS, you will need to do the following:

Download Visual Studio for C++: https://visualstudio.microsoft.com/vs/features/cplusplus/
Download additional documents specific to Windows in the Windows_Specific folder. Move these files into your [ale] folder.
Run the following commands in MATLAB beforehand:
```
mex -setup
mex 'tfceMex.c'
mex 'tfceMex_pthread.c'
```
You are now all set for running the ALE scripts! One other caveat with Windows is you will need a way to unzip .gz files. We recommend the program WinRAR for this.

Conduct Meta-analysis

Main Analyses

Read data

To run the analyses, first make sure the coordinate data is read in to the spreadsheet by running the line ale_inputCoords('data/aberrant_20200113.xls'). This reads the coordinate data for the main analyses (aberrant activity) located in data/aberrant_20200113.xlsx and converts it into a .mat format for analyses. Next, run the line ale_inputCoords('data/painCoords_20200113.xls'); to read the coordinate data for subset analyses of signed differences. This coordinate data is located in data/painCoord_20200113.xls. These coordinate data will now be read in .mat format for analyses and are located in ale/DataMatlab.

ROI-based analyses set up

To conduct ROI-based analyses, fist download the Xu et al., 2020 meta-analytic map from Neurovault as Volume: https://neurovault.org/collections/AJKXTUXX/images/359250/. On your terminal, binarize this image using the line fslmaths "pain_thresh_cFWE05.nii.gz" -bin "healthyPainBin.nii.gz". This outputs a binarized nifti named healthyPainBin.nii.gz. Unzip this file and move healthyPainBin.nii into the ale folder.

Run analyses

After, run the line ale_inputCoords(data/painCoords_20200113.xls) This reads in the data for signed differences between patients and controls (i.e., separate sub-analyses of experiments using “patients > controls” contrast and experiments using “controls > patients” contrast), located in data/painCoords_20200113.xlsx.

Next, run the line ale_estimateALE('contrasts/pain_20200113.xlsx'); to run all analyses. This line reads the spreadsheet contrasts/pain_20200113.xlsx and runs the contrasts outlined there. ROI-based analyses will also be conducted with this line.

The total runtime on the computer that ran the analyses locally (Macbook Pro 2018) is around 7 hours. That is the typical expected time on newer laptops.

Perceptual Ratings Subset

To run sub-analyses looking only at experiments matching pain stimuli between patients and controls with similar painfulness ratings, run the line ale_inputCoords('data/perceptualData_2020-01-21.xls'); to read the relevant coordinate data.

Next, run the line ale_estimateALE('contrasts/perceptual_20200121.xlsx'); to run the relevant contrasts.

Interpret Results

The results are now located in a newly generated folder ale/ALE. Inside this folder, there is a folder named Results. This folder contains the thresholded meta-analytic maps. There is another folder called Contribution, which contains any significant coordinate results and % contribution from each experiment. Z-scored unthresholded are located in VolumesZ while foci images are located in Foci. Results from ROI-based analyses are located in ROIresults. These results are saved as .pdf files.

Generate figures

All surface figures were generated using BrainNet Viewer. To use BrainNet, simply type BrainNet into the command line in MATLAB.

Unthresholded maps were loaded as the mapping image while BrainNet_ICBM152.nv was loaded as the surface file. Images were thresholded to positive values only for display. This is done by going to BrainNet_Option > Volume > Display: Positive only.

Axial view of images were generated using fslview. Images were thresholded to positive values only.

Distribution of foci

Surface figures of foci distribution were generated using BrainNet. Before creating these figures, you need a .node file for BrainNet to read in. This can be generated by creating a .csv file in the main folder and running the tal2mni_coordinates_20200131.m script. This script reads the file coordinates_20200131.csv file, which is just the coordinates used in the meta-analysis copied as a separate spreadsheet. Next, it converts any Tal coordinates to MNI and then outputs a .csv file with all coordinates in MNI.

Once you have the outputted spreadsheet, create three new columns next to the MNI coordinates. In the first new column, type 1 for all coordinates that are “patient > control” contrast and type 0 for all coordinates that are “control > patient” contrast (this information is under the header “contrast”). In the second new column, type 1. In the third new column, type -.

After you’re done editing the spreadsheet, copy and paste the MNI coordinates under the columns mni_x, mni_y, and mni_z into a text editor as well as the three new columns generated. If you use TextEditor in MacOS, enter command+shift+t first to make it unformatted text editing. Delete the column header (i.e., the first row that contains the column names mni_x, mni_y, and mni_z), and then save this file as coordinates.node. Please note the coordinates displayed in the paper do not display coordinates for cerebellum.

Open BrainNet in MATLAB. Next, choose BrainNet_ICBM152.nv as the surface figure and choose coordinates.node for Data file (node). Under “Node” in Options, go to “Color” and choose the option “Modular”. In “Modular”, click “More” and select the color red as Module1 and blue as Module2. Click “OK” and generate the figure. Please note

Xu_fMRIChronicPain