What was new in previous releases?#

Version 0.7 (2024-06-03)#

This version adds a new multivariate connectivity method to the spectral connectivity analysis functions, an example explaining when certain connectivity methods should be used, and a function for simulating connectivity. Computation time has also been improved in some cases, including a bug fix for parallel processing.

Enhancements#

Add support for a new multivariate connectivity method (canonical coherence; cacoh) in mne_connectivity.spectral_connectivity_epochs() and mne_connectivity.spectral_connectivity_time(), by Thomas Binns and Mohammad Orabe and Mina Jamshidi (#163).
Add a new example for comparing use-cases of different coherency-based methods, by Thomas Binns and Mohammad Orabe (#163).
Add a new function mne_connectivity.make_signals_in_freq_bands() for simulating connectivity between signals, by Thomas Binns and Adam Li (#173).

Bug#

Fix bug where n_jobs=-1 would not be converted to the CPU count in mne_connectivity.spectral_connectivity_epochs(), by Thomas Binns (#177).

Authors#

Version 0.6 (2023-12-06)#

This version introduces new functionality for computing multivariate connectivity methods, championed primarily by Thomas Binns. The spectral_connectivity_time and spectral_connectivity_epochs functions now supports multivariate connectivity methods.

Enhancements#

Add the option to set the number of connections plotted in mne_connectivity.viz.plot_sensors_connectivity() by Qianliang Li (#133).
Allow setting colormap via new parameter cmap in mne_connectivity.viz.plot_sensors_connectivity() by Daniel McCloy (#141).
Add support for multivariate connectivity methods in mne_connectivity.spectral_connectivity_epochs() and mne_connectivity.spectral_connectivity_time() by Thomas Binns and Tien Nguyen and Richard Köhler (#138) and (#142).

Version 0.5 (2023-01-13)#

This version has several improvements in the spectral connectivity analysis and some bug fixes. Functionality now requires mne>=1.3 for MNE-Python as a minimum version requirement.

This version has major changes in mne_connectivity.spectral_connectivity_time(). Several bugs are fixed, and the function now computes static connectivity over time, as opposed to static connectivity over trials computed by mne_connectivity.spectral_connectivity_epochs().

Enhancements#

Add the PLI and wPLI methods in mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#104).
Improve the documentation of mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#104).
Add the option to average connectivity across epochs and frequencies in mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#104).
Select multitaper frequencies automatically in mne_connectivity.spectral_connectivity_time() similarly to mne_connectivity.spectral_connectivity_epochs() by Santeri Ruuskanen (#104).
Add the ciPLV method in mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#115).
Add the option to use the edges of each epoch as padding in mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#115).

Bug#

When using the multitaper mode in mne_connectivity.spectral_connectivity_time(), average CSD over tapers instead of the complex signal by Santeri Ruuskanen (#104).
Average over time when computing connectivity measures in mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#104).
Fix support for multiple connectivity methods in calls to mne_connectivity.spectral_connectivity_time() by Santeri Ruuskanen (#104).
Fix bug with the indices parameter in mne_connectivity.spectral_connectivity_time(), the behavior is now as expected by Santeri Ruuskanen (#104).
Fix bug with parallel computation in mne_connectivity.spectral_connectivity_time(), add instructions for memory mapping in doc by Santeri Ruuskanen (#104).
Allow loading files that don’t have event_id_keys and event_id_values defined, by Daniel McCloy (#110)
Fix handling of the verbose argument by spectral_connectivity_epochs(), spectral_connectivity_time(), vector_auto_regression() by Sam Steingold (#111)

API#

Streamline the API of mne_connectivity.spectral_connectivity_time() with mne_connectivity.spectral_connectivity_epochs() by Santeri Ruuskanen (#104).
The sfreq parameter is now required for numpy array inputs to spectral_connectivity_epochs(), by Adam Li (#119)

Authors#

Version 0.4 (2022-10-05)#

There are a few enhancements. Notably, the dPLI method was added to mne_connectivity.spectral_connectivity_epochs(). There are some bug fixes to the underlying API to make compatible with MNE-Python v1.2+.

Enhancements#

Add node_height to mne_connectivity.viz.plot_connectivity_circle() and enable passing a polar ax, by Alex Rockhill (#88)
Add directed phase lag index (dPLI) as a method in mne_connectivity.spectral_connectivity_epochs() with a corresponding example by Kenji Marshall (#79)

Bug#

Fix the output of mne_connectivity.spectral_connectivity_epochs() when faverage=True, allowing one to save the Connectivity object, by Adam Li and Szonja Weigl (#91)
Fix the incompatibility of dimensions of frequencies in the creation of EpochSpectroTemporalConnectivity object in mne_connectivity.spectral_connectivity_time() by providing the frequencies of interest into the object, rather than the frequencies used in the time-frequency decomposition by Adam Li and Sezan Mert (#98)

Authors#

Version 0.3 (2022-03-01)#

This version has bug fixes minor improvements in certain functions. A big change is the renaming of functions spectral_connectivity to spectral_connectivity_epochs, which makes it explicit that the function operates over Epochs, rather then time. Importantly, we also provide a conda installation now.

Enhancements#

Adding symmetric orthogonalization via mne_connectivity.symmetric_orth(), by Eric Larson (#36)
Improved RAM usage for mne_connectivity.vector_auto_regression() by leveraging code from statsmodels, by Adam Li (#46)
Added mne_connectivity.select_order() for helping to select VAR order using information criterion, by Adam Li (#46)
All connectivity functions retain events, event_id and metadata from mne.Epochs objects as input and is stored as part of the connectivity object, by Adam Li (#58)
Add spectral connectivity over time function mne_connectivity.spectral_connectivity_time(), by Adam Li (#67)
Add conda installation, by Adam Li and Richard Höchenberger (#81)

Bug#

Fixed bug when saving connectivity with n_jobs greater than 1 from mne_connectivity.spectral_connectivity_epochs(), by Adam Li (#43)
Fixed bug to allow saving complex data connectivity, by Adam Li (#43)
Fixed bug to keep label orientation upright in mne_connectivity.viz.plot_connectivity_circle(), by Alexander Kroner (#60)

API#

Added h5netcdf as a requirement for saving connectivity data, by Adam Li (#43)
Changed keyword argument model_order in mne_connectivity.vector_auto_regression() to lags to more align with statsmodels API, by Adam Li (#47)
Add pandas as a requirement for dealing with metadata associated from the original Epochs file, by Adam Li (#58)
Rename mne_connectivity.spectral_connectivity to mne_connectivity.spectral_connectivity_epochs(), by Adam Li (#69)

Authors#

People who contributed to this release (in alphabetical order):

Version 0.2 (2021-09-07)#

In this version, we return the relevant Connectivity class from each of the connectivity estimation functions. These internally use xarray to represent the connectivity data. One can easily get the v0.1 numpy array by doing conn.get_data(), which will get exactly the same output as one got in v0.1 running any of the connectivity functions.

Changelog#

Adding Connectivity, TemporalConnectivity, SpectralConnectivity and SpectroTemporalConnectivity as a data structure to hold connectivity data, by Adam Li (#6)
Adding EpochConnectivity, EpochTemporalConnectivity, EpochSpectralConnectivity and EpochSpectroTemporalConnectivity as a data structure to hold connectivity data over Epochs, by Adam Li (#6)
indices argument in Connectivity classes can now be symmetric, allowing for memory-efficient storage of symmetric connectivity, by Adam Li (#20)
New function save in Connectivity classes along with read_connectivity() can now be used to write and read Connectivity data as netCDF files, by Adam Li (#20)
New function vector_auto_regression() to compute dynamic connectivity vector auto-regressive (VAR) model, by Adam Li (#23)

API#

envelope_correlation(), spectral_connectivity, and phase_slope_index() all return _Connectivity containers now, by Adam Li (#6)
Added xarray as a dependency where all connectivity containers are now underlying xarrays, by Adam Li (#6)
The combine argument in envelope_correlation() was removed, and now all Epoch Connectivity classes have a combine class function, by Adam Li (#20)

Authors#

People who contributed to this release (in alphabetical order):

Version 0.1 (2021-06-25)#

Changes when mne-connectivity was part of MNE-Python#

In July, 2021, mne.connectivity submodule was ported over from the MNE-Python repo into this repository, mne-connectivity as v0.1. Starting v0.24 of MNE-Python, that sub-module will be deprecated and development will move over into this repository. Starting v0.25 of MNE-Python, mne.connectivity will completely be removed.

Authors#

People who contributed to this release (in alphabetical order):

Adam Li