Data from: Rhythmicity of neuronal oscillations delineates their cortical and spectral architecture
Data files
Mar 15, 2024 version files 10 GB
-
meso_submission_data_final.zip
-
README.md
Abstract
Neuronal oscillations are commonly analyzed with power spectral methods that quantify signal amplitude, but not rhythmicity or 'oscillatoriness' per se. Here we introduce a new approach, the phase-autocorrelation function (pACF), for direct quantification of rhythmicity. We applied pACF to human intracerebral stereo-electroencephalography (SEEG) and magnetoencephalography (MEG) data and uncovered a spectrally and anatomically fine-grained cortical architecture in the rhythmicity of single- and multi-frequency neuronal oscillations. Evidencing the functional significance of rhythmicity, we found it to be a prerequisite for long-range synchronization in resting-state networks and to be dynamically modulated during event-related processing. We also extended the pACF approach to measure 'burstiness' of oscillatory processes and characterized regions with stable and bursty oscillations. These findings show that rhythmicity is double-dissociable from amplitude and constitutes a functionally relevant and dynamic characteristic of neuronal oscillations.
README
Usage notes
The data consists of Python numpy and pickle files and is grouped in the next directories:
- anatomy * brain_anatomy.pickle - a file with brain surfaces required for visualization of brain maps
- MEG\ Common MEG derivatives that are used across multiple figures. In includes pACF and PSD brain maps for the MEG data, wPLI for the resting-state cohort, surrogate pACF level, parcel names mapping between yeo7 and yeo17 parcellations. * 7_to_17_400.csv \ A table with parcel names mapping between yeo7 and yeo17 parcellations * meg_noise_level.npy \ pACF noise level for MEG data * meg_pac_results_full.pickle \ pACF brain map for the resting-state MEG cohort * meg_pac_with_statistics_wpli.pickle \ pACF with wPLI and PSD statistics for the MEG cohort
- SEEG\ Common SEEG derivatives that are used across multiple figures. In includes pACF brain maps for the SEEG data, surrogate pACF level, number of contacts per parcel and the parcellation adjacency matrix. * adjacency_matrix.npy - binary adjacency matrix for the SEEG parcellation * cohort_seeg_pac_significant.npy - significant pACF values for the SEEG cohort * cohort_seeg_pac_values.npy - pACF values for the SEEG cohort * counter_known.npy - number of contacs per parcel * noise_pacf_freqwise.npy - surrogate-level cohort pACF values for the SEEG cohort * Schaefer2018_400Parcels_17Networks_order_LUT - LUT for the SEEG parcellation * seeg_noise_level.npy - pACF noise level for SEEG data
- figure_1\ Data files specific for reproduction of the first figure including pACF statistics for different kinds of simulated data. * corr_data.pickle - autocorrelation function values for amplitude-normed and original narrow-band signal * noise_exp_data.pickle - pACF properties for the noise with different power-law exponentials * non_stat_data.pickle - pACF properties for the oscillaty signal with different stationarity properties * pac_data.pickle - pACF properties for signals with varied rhythmicity and power
- figure_2\ Data files specific for reproduction of the second figure including individual PSD, pACF curves and lifetimes. * example_subject - pACF curve values for an example subject * figure_psd_data - PSD values for an example subject * target_subject - pACF lifetime values for an example subject * noise_level - pACF surrogate level for an example subject * noise_mean - mean surrogate pACF for an example subject
- figure_3\ Data files specific for reproduction of the third figure including surrogate level pACF correlations between different frequencies, spectrum of significant PSD values. * cohort_seeg_psd_significant - brain map of significant PSD values for the SEEG cohort * meg_pac_data.pickle - MEG pACF brainmap * surr_freq_corr_values.npy - distribution of surrogate frequency-frequency correlations
- figure_4\ Data files specific for reproduction of the fourth figure including brain maps of peak frequencies. * seeg_cortical_peaks_boot.npy - distribution of peak frequencies per parcel * seeg_peaks_map.pickle
- figure_5\ Data files specific for reproduction of the fifth figure including correlations between pACF lifetime, PSD and Amplitude Spectrum as a function of frequency for SEEG and MEG data. * psd_correlations_eyes.pickle - correlation of PSD spectra with amplitude and pACF lifetime values for the MEG data * seeg_data.pickle - correlation of PSD spectra with amplitude and pACF lifetime values for the SEEG data
- figure_6\ Data files specific for reproduction of the sixth figure including Functional Connectivity (FC) data for the simulated data (PLV), SEEG (PLV) and MEG (wPLI), correlations between their pACF lifetimes and FC and surrogate data. * kuramoto_pac_plv.pickle - single-simulation pACF and PLV values * kuramoto_pac_plv_cohort.pickle - pACF and PLV values across multiple simulations * kuramoto_plv_vs_pac.pickle - binned PLV as a function of pACF for the simulation data * seeg_data.pickle - pACF and PLV correlation for the SEEG data * surr_levels.pickle - surrogate levels for the correlation and predictivity
- figure_7\ Data files specific for reproduction of the seventh figure including stability index brain maps for SEEG and MEG data, surrogate level and frequency communities for SEEG and MEG data separately. * cohort_seeg_burst_known.npy - stability index for the SEEG cohort * burst_coeff_example.pickle - simulated timeseries with burst-like and stable behaviour * meg_burst_data.pickle - stability index for the MEG cohort * meg_communities_new - frequency communities for the MEG cohort * meg_data_nsamples.pickle - number of samples in MEG recordings * meg_noise_level.npy - noise-level for the MEG data * noise_pacf_by_ncycles.npy - surrogate pACF level as a function of number of cycles * seeg_communities_new - frequency communities for the SEEG cohort * seeg_data_nsamples.pickle - number of samples in SEEG cohort * seeg_new_names_yeo17_400 - parcel names for the SEEG data * yeo17_400_adjacency_matrix.npy - adjacency matrix for the Schaefer yeo17 parcellation with 400 parcels
- figure_8\ Data files specific for reproduction of the eighth figure including TSDT TFR pACF, amplitude and phase-reset responses and their surrogate levels. * tsdt_data.pickle - TSDT brain maps for different kinds of responses * tsdt_surr_level_parcelwise.npy - surrogate level for the TSDT pACF response * tsdt_surr_level_parcelwise_amp.npy - surrogate level for the TSDT amplitude response * tsdt_surr_level_parcelwise_evoked.npy - surrogate level for the TSDT phase-reset response
Code/Software
All the code is written in python + several CUDA functionals and either Jupyter notebooks (to reproduce the paper figures) or Python scripts (to process raw data files). The code is open-source and can be found on github
Methods
We assembled a large database of human intracerebral stereoelectroencephalography (SEEG, N=61) and magnetoencephalography (MEG, N=52) recordings. We applied the Phase Autocorrelation Function (pACF) for such recordings to obtain the rhytmicity map of human cortex for both modalities. We also applied the time-resolved pACF to analyze the event-related visual Threshold-Stimulus Detection Task (TSDT) and build maps of responses for it. To compare other methods to the pACF we also computed Power Spectral Density (PSD) and Wavelet Amplitude Spectra for the resting-state recordings alongside with evoked and phase-reset responses for the TSDT data.