Version 1.2.0 (2022-10-11)

Enhancements

• Add vmin and vmax parameters to mne.Evoked.animate_topomap() (#11073 by new contributor Mats van Es)

• Enable common average reference projection for ECoG, sEEG, and DBS data (optionally jointly with joint=True) in set_eeg_reference() and related functions/methods (#10656 by new contributor Moritz Gerster and Eric Larson)

• The trans parameter in mne.make_field_map() now accepts a Path object, and uses standardised loading logic (#10784 by new contributor Andrew Quinn)

• Add support for computing forward solutions using OpenMEEG (#11011 by Alex Gramfort and Eric Larson)

• EEGLAB files (saved as MAT versions less than v7.3) can now be imported with mne.io.read_raw_eeglab() without the optional dependency pymatreader (#11006 by Clemens Brunner)

• Add eight source estimate comparison metrics to new submodule mne.simulation.metrics (#6233 by Kostiantyn Maksymenko and Alex Gramfort)

• Add mne.time_frequency.csd_tfr() to compute cross-spectral density from mne.time_frequency.EpochsTFR (#10986 by Alex Rockhill)

• Improve repr() for mne.minimum_norm.InverseOperator when loose orientation is used (#11048 by Eric Larson)

• Add support for vector and volume source estimates in mne.minimum_norm.get_point_spread() and mne.minimum_norm.get_cross_talk() (#8639, #11083 by Olaf Hauk, Marijn van Vliet, Alex Gramfort, and Eric Larson)

• mne.Epochs.plot_psd_topomap() now suppresses redundant colorbars when vlim='joint' (#11051 by Daniel McCloy)

• Add starting_affine keyword argument to mne.transforms.compute_volume_registration() to initialize an alignment with an affine (#11020 by Alex Rockhill)

• Add HTML representation for Evoked in Jupyter Notebooks (#11075 by Valerii Chirkov and Andrew Quinn)

• Add support for temperature and gsr (galvanic skin response, i.e., electrodermal activity) channel types (#11090, #11108 by Eric Larson and Richard Höchenberger)

• Allow mne.beamformer.make_dics() to take pick_ori='vector' to compute vector source estimates (#10980 by Alex Rockhill)

• Add units parameter to mne.io.read_raw_edf() in case units are missing from the file (#11099 by Alex Gramfort)

• Add on_missing functionality to all of our classes that have a drop_channels method, to control what happens when channel names are not in the object (#11077 by Andrew Quinn)

• Improve interpolation of bridged electrodes with interpolate_bridged_electrodes (#11094 by Mathieu Scheltienne)

• Add mne.minimum_norm.apply_inverse_tfr_epochs() to apply inverse methods to time-frequency resolved epochs (#11095 by Alex Rockhill)

• Add support for multiple probes via multiple CSV files passed to mne.io.read_raw_hitachi() (#11186 by Eric Larson)

• Add mne.chpi.get_active_chpi() to retrieve the number of active hpi coils for each time point (#11122 by Eduard Ort)

• Add example of how to obtain time-frequency decomposition using narrow bandpass Hilbert transforms to Time-frequency on simulated data (Multitaper vs. Morlet vs. Stockwell vs. Hilbert) (#11116 by Alex Rockhill)

• Add == and != comparison between mne.Projection objects (#11147 by Mathieu Scheltienne)

• Parse automatically temperature channel with mne.io.read_raw_edf() (#11150 by Eric Larson and Alex Gramfort)

• Add unscale option to mne.head_to_mri() to facilitate working with surrogate MRI data (#11185 by Eric Larson)

• Add encoding parameter to mne.io.read_raw_edf() and mne.io.read_raw_bdf() to support custom (non-UTF8) annotation channel encodings (#11154 by Clemens Brunner)

• mne.preprocessing.ICA gained a new method, get_explained_variance_ratio(), that allows the retrieval of the proportion of variance explained by ICA components (#11141 by Richard Höchenberger)

• Add on_baseline to mne.preprocessing.ICA.apply(), mne.preprocessing.ICA.plot_overlay(), and mne.viz.plot_ica_overlay() to allow reapplying baseline correction after applying ICA (#11232 by Eric Larson)

• Add config option MNE_REPR_HTML to disable HTML repr in notebook environments (#11159 by Clemens Brunner)

Bugs

• Fix bug in mne.io.Raw.crop() where argument include_tmax was not considered in checking tmax in bounds (#11196 by Lukas Gemein)

• Fix bug in mne.io.read_raw_eeglab() where unlabeled fiducials caused reading errors (#11074 by new contributor Sebastiaan Mathot)

• Fix bug in mne.time_frequency.read_csd() that returned projs as a list of dict instead of mne.Projection (#11072 by new contributor Chetan Gohil)

• Fix bug in mne.decoding.TimeFrequency() that prevented cloning if constructor arguments were modified (#11004 by new contributor Daniel Carlström Schad)

• Fix bug in repr of decoding estimator classes (e.g., mne.decoding.SlidingEstimator) where the parameters where not stored in the result string but instead just printed to sys.stdout (#11190 by new contributor Dominik Wetzel)

• Document height and weight keys of subject_info entry in mne.Info (#11019 by new contributor Sena Er)

• Fix bug in mne.viz.Brain constructor where the first argument was named subject_id instead of subject (#11049 by Eric Larson)

• Fix bug in mne coreg where the MEG helmet position was not updated during ICP fitting (#11084 by Eric Larson)

• Fix bug in read_raw_curry(), and read_raw_cnt() where digitization points were not read properly (#11145 by Eric Larson)

• Fix bug in mne.minimum_norm.apply_inverse_epochs() where the average EEG projector was not checked properly (#11182 by Eric Larson)

• Fix bug in mne.viz.plot_filter() when plotting filters created using output='ba' mode with compensation turned on. (#11040 by Marian Dovgialo)

• Fix bugs in documentation of surface SourceSpaces (#11171 by Eric Larson)

• Fix bug in mne.viz.plot_compare_evokeds() where automatic legend labels could be excessively long; they are now abbreviated with ... when necessary (#11181 by Eric Larson)

• Fix bugs with verbose='error' not being used properly and allowing warnings through (#11193 by Eric Larson)

• Fix bug in mne.io.read_raw_bti() where EEG, EMG, and H/VEOG channels were not detected properly, and many non-ECG channels were called ECG. The logic has been improved, and any channels of unknown type are now labeled as misc (#11102 by Eric Larson)

• Fix bug in mne.viz.plot_topomap() when providing sphere="eeglab" (#11081 by Mathieu Scheltienne)

• Fix bug in mne.Dipole.to_mri() where MRI RAS rather than MRI surface RAS was returned (#11185 by Eric Larson)

• Fix bug in epochs.save where the verbose parameter defaulted to True instead of None (#11191 by Eric Larson)

• The string and HTML representation of mne.preprocessing.ICA reported incorrect values for the explained variance. This information has been removed from the representations, and should instead be retrieved via the new get_explained_variance_ratio() method (#11141 by Richard Höchenberger)

• Fix bug in mne.Evoked.plot() and related methods where a np.nan location value in any channel causes spatial colours to fail (#6870 by Simeon Wong)

• Fix bug in mne.preprocessing.ICA.find_bads_muscle() where epochs caused an error when passed as the inst (#11197 by Alex Rockhill)

• Fix bug in readers where EEG coordinates were assumed to be in head coordinates but no fiducial points were present. Estimated fiducial locations will now be added automatically to reflect the assumption of locations being in the head coordinate frame (#11212 by Stefan Appelhoff and Eric Larson)

• The duration of raw data sometimes wasn’t displayed correctly in Jupyter notebooks by omitting fractions of a second. We now always round up to the next full second so a duration of less than 1 second will not be displayed as a duration of zero anymore (#11203 by Richard Höchenberger)

• Fix bug in mne.Report.add_ica() and where baselines were not reapplied to the data when inst is Epochs or Evoked (#11232 by Eric Larson)

API changes

• In meth:mne.Evoked.plot, the default value of the spatial_colors parameter has been changed to 'auto', which will use spatial colors if channel locations are available (#11201 by new contributor Hüseyin Orkun Elmas and Daniel McCloy)

• Starting with this release we now follow the Python convention of using FutureWarning instead of DeprecationWarning to signal user-facing changes to our API (#11120 by Daniel McCloy)

• The names parameter of mne.viz.plot_arrowmap() and mne.viz.plot_regression_weights() has been deprecated; sensor names will be automatically drawn from the info_from or model parameter (respectively), and can be hidden, shown, or altered via the show_names parameter (#11123 by Daniel McCloy)

• The bands parameter of mne.Epochs.plot_psd_topomap() now accepts dict input; legacy tuple input is supported, but discouraged for new code (#11050 by Daniel McCloy)

• The mne.head_to_mri() new function parameter kind default will change from 'ras' to 'mri' (#11185 by Eric Larson)

• The show_toolbar argument to mne.viz.Brain is being removed by deprecation (#11049 by Eric Larson)

• New classes Spectrum and EpochsSpectrum, created via new methods Raw.compute_psd(), Epochs.compute_psd(), and Evoked.compute_psd() (#10184 by Daniel McCloy)

• Functions/methods for plotting topomaps now have more consistent signatures; most have gained parameters for finer control of the topomap image and/or colorbar. See docstrings for details (#11123 by Daniel McCloy)

• In mne.viz.plot_topomap(), passing axes=None will now create a new Figure instead of plotting into the current axes (#11123 by Daniel McCloy)

• In mne.viz.plot_topomap(), the parameter show_names has been deprecated, use the names parameter instead (#11123 by Daniel McCloy)

• In mne.viz.plot_topomap(), plot_tfr_topomap(), plot_evoked_topomap(), and related class methods mne.Covariance.plot_topomap(), mne.Evoked.plot_topomap(), and mne.time_frequency.AverageTFR.plot_topomap(): the parameters vmin and vmax are deprecated in favor of vlim, for consistency with other plot_topomap functions and methods (#11123 by Daniel McCloy)

• In mne.viz.plot_tfr_topomap(), plot_evoked_topomap(), and related class methods mne.Covariance.plot_topomap(), mne.Evoked.plot_topomap(), and mne.time_frequency.AverageTFR.plot_topomap(): the title parameter has been deprecated; use matplotlib’s fig.suptitle() instead (#11123 by Daniel McCloy)

• In functions/methods for plotting topomaps, the outlines='skirt' option is deprecated; use sphere='eeglab' for similar functionality (#11123 by Daniel McCloy)

• The mne.epochs.add_channels_epochs function has been deprecated in favor of epochs.add_channels (#11180 by Eric Larson)

• The PSD functions that operate on Raw/Epochs/Evoked instances (mne.time_frequency.psd_welch and mne.time_frequency.psd_multitaper) are deprecated; for equivalent functionality create Spectrum or EpochsSpectrum objects instead and then run spectrum.get_data(return_freqs=True) (#10184 by Daniel McCloy)

• Added new class mne.preprocessing.EOGRegression to allow more flexibility when using regression to reduce EOG artifacts (#11046 by Marijn van Vliet)

• New parameter exclude added to mne.preprocessing.regress_artifact() to prevent regression from being applied to certain channels (#11046 by Marijn van Vliet)

Authors

• Alex Rockhill

• Alexandre Gramfort

• Andrew Quinn+

• Chetan Gohil+

• Christian Brodbeck

• Clemens Brunner

• Daniel Carlström Schad+

• Daniel McCloy

• Dominik Wetzel+

• Eduard Ort

• Eric Larson

• Hüseyin Orkun Elmas+

• Jack Zhang

• Joan Massich

• Julius Welzel+

• Kostiantyn Maksymenko

• Lukas Gemein

• Marian Dovgialo

• Marijn van Vliet

• Mathieu Scheltienne

• Mats van Es+

• Moritz Gerster+

• Olaf Hauk

• Richard Höchenberger

• Sebastiaan Mathot+

• Sena Er+

• Simeon Wong

• Stefan Appelhoff

• Valerii Chirkov

• luzpaz+