Timescales of spontaneous fMRI fluctuations relate to structural connectivity in the brain

This repository contains code to reproduce the key figures from our publication:

• 📙 Fallon et al. (2020), Network Neuroscience. Timescales of spontaneous fMRI fluctuations relate to structural connectivity in the brain.

Dependencies

• Some code (for computing timescales) uses CO_AutoCorrShape and dependent functions in hctsa (v1.01 was used for published results).
• Some functions, load_nii, require having tools for reading NIfTI images (e.g., the NIfTI toolbox) installed and in the Matlab path.

Data

Data are available from this Zenodo repository and should be placed in the Data directory as follows:

• Subject info: Data/subs100.mat.
  Contains information about all subjects analyzed.
• Structural connectomes: Data/connectome/
  Contains structural connectivity data for the three parcellations investigated here.
• Regional time series: Data/rsfMRI/.
  Contains a cfg.mat file for all subjects.
• Region volumes: Data/volume/.
  Contains volume info for all ROIs in each of the three parcellations investigated.
• Results of hctsa analysis: Data/hctsa_stats.mat.
• Surface for surface plotting: Data/fsaverage_surface_data.mat.

Analysis code

Add paths to all subdirectories by running startup.

Plots of data for the schematic

Produce data for schematic figure (Fig. 1):

dataPlotsForSchematic()

(Also outputs some surface-space plots used in Fig. 2D)

Relative low-frequency power as a function of node strength (+ partial correction):

Produces Fig. 2A:

params = GiveMeDefaultParams('DK');
PlotNSScatter(params,'RLFP')

This outputs several figures and correlation statistics to the command-line:

Description	Output
Node strength scatter (correlation and p-value in title)
Fig. 2A: Residuals from region-volume variation (correlation and p-value in title)
Labeling of data points by region ID
Volume scatter

These results can be re-run for 'timescale' or 'fALFF' instead of 'RLFP'.

You can also run with different parcellations by modifying the corresponding element of the params structure.
For example, to produce Fig. 2C: params = GiveMeDefaultParams('cust200');.

Plot power spectral density curves for selected regions

Produces Fig. 2B:

PSD_plot()

Inter-individual differences in correlations

Produces Fig. 3:

InterIndividual()

Comparison of selected feature to others from hctsa

hctsaCorr()

Fig 4:

Also the raw distribution (without absolute value or taking residuals from volume):