Version 0.20 (2020-03-27)

Changelog

• Improved mne.viz.plot_epochs() to label epoch counts starting from 0, by Sophie Herbst

• Add minimum_norm.apply_inverse_cov() to compute static power by applying inverse solutions to a data covariance matrix by Denis Engemann, Luke Bloy, and Eric Larson

• Add mne.minimum_norm.resolution_metrics() to compute various resolution metrics for inverse solutions, by Olaf Hauk

• Add current source density mne.preprocessing.compute_current_source_density() to compute the surface Laplacian in order to reduce volume conduction in data by Alex Rockhill

• Add mne.chpi.extract_chpi_locs_ctf() to extract cHPI coil locations from CTF data by Luke Bloy

• Add mne.chpi.compute_chpi_amplitudes(), mne.chpi.compute_chpi_locs(), and mne.chpi.compute_head_pos() to compute head positions from cHPI coil locations by Eric Larson and Luke Bloy

• Add allow_line_only option to mne.chpi.filter_chpi() to allow filtering line frequencies only in files that do not have cHPI information by Eric Larson

• Add mne.io.Raw.set_meas_date() by Eric Larson

• Add mne.Epochs.as_type() to allow remapping data in MEG channels to virtual magnetometer or gradiometer channels by Sophie Herbst and Alex Gramfort

• Add copy parameter to mne.Epochs.iter_evoked() by Alex Gramfort

• Add command mne setup_source_space to quickly set up bilateral hemisphere surface-based source space with subsampling by Victor Ferat.

• Add command mne sys_info to print system information by Eric Larson

• Add function mne.make_fixed_length_epochs() to segment raw into fixed length epochs by Mohammad Daneshzand

• Add support for computing patch information only in surface source space creation with add_dist='patch' in mne.setup_source_space() and dist_limit=0 in mne.add_source_space_distances() by Eric Larson

• Add mne.Label.restrict to restrict a label to vertices within a source space by Eric Larson

• Add support for passing a destination source space src_to in mne.compute_source_morph() to ensure morphing for multiple subjects results in consistent STCs, by Eric Larson

• Add support for smooth='nearest' for surfaces in mne.compute_source_morph() by Eric Larson

• Add support for plotting fNIRS channels in mne.viz.plot_alignment(), mne.viz.plot_evoked_topo() and mne.Evoked.animate_topomap() by Eric Larson and Robert Luke

• Add command line tool mne anonymize for anonymizing raw fiff files by Luke Bloy

• Add mne.preprocessing.find_bad_channels_maxwell() for automatic bad channel detection via Maxwell filtering by Eric Larson

• Add support to mne.io.anonymize_info() to allow time offset to be applied to dates by Luke Bloy

• Add support for computing resolution matrix to get point spread functions (PSF) and cross-talk functions (CTF) in mne.minimum_norm.make_inverse_resolution_matrix(), mne.beamformer.make_lcmv_resolution_matrix(), mne.minimum_norm.get_cross_talk(), mne.minimum_norm.get_point_spread() by Olaf Hauk

• Add mne.preprocessing.read_ica_eeglab() to read EEGLAB ICA decompositions by Christian O’Reilly

• Add keyboard functionality to interactive colorbar plotting TFRs by Stefan Repplinger

• Add the ability to return_event_id in mne.read_events() for use with MNE-C produced -annot.fif files, by Eric Larson

• Add option to mne.connectivity.spectral_connectivity to compute corrected imaginary PLV by Adonay Nunes

• Add mne.SourceEstimate.estimate_snr() to estimate source-space SNR, by Kaisu Lankinen and Padma Sundaram

• Add option to specify the coordinate frame in mne.channels.read_custom_montage() by Eric Larson

• Add option to use match_case=False to mne.io.Raw.set_montage() and related functions by Eric Larson

• Add reader for NIRx data in mne.io.read_raw_nirx() by Robert Luke

• Add function to convert NIRS data to optical density mne.preprocessing.nirs.optical_density() by Robert Luke

• Add function to convert NIRS data to haemoglobin concentration mne.preprocessing.nirs.beer_lambert_law() by Robert Luke

• Add fig argument to mne.viz.plot_evoked_field() by Eric Larson

• Add functions to calculate spatial information of NIRS channels mne.preprocessing.nirs.source_detector_distances() and mne.preprocessing.nirs.short_channels() by Robert Luke

• Add reader for *.dat electrode position files mne.channels.read_dig_dat() by Christian Brodbeck

• Improved mne.viz.plot_events() to always show event counts by Eric Larson

• Improved LIMO Dataset usage and example for usage of mne.stats.linear_regression() by Jose Alanis

• Add support for reduce_rank=True for vector beamformers by Eric Larson

• Add method='sphere' support for using Freesurfer spherical inflation in mne.decimate_surface() to facilitate mesh downsampling for use with mne.make_bem_model() by Eric Larson

• Speed up mne.beamformer.make_lcmv() and mne.beamformer.make_dics() calculations by vectorizing linear algebra calls by Dmitrii Altukhov and Eric Larson

• Speed up mne.make_forward_solution() using Numba, by Eric Larson

• Speed up mne.io.read_raw_fif() data reading when the recording is long and there are many data tags repeatedly accessed, by Eric Larson

• For KIT systems without built-in layout, mne.channels.find_layout() now falls back on an automatically generated layout, by Christian Brodbeck

• mne.Epochs.plot() now takes a epochs_colors parameter to color specific epoch segments by Mainak Jas

• Add command mne setup_forward_model to quickly create a BEM model for a subject by Victor Ferat.

• Add command mne prepare_bem_model to quickly create a BEM solution using the linear collocation approach by Victor Ferat.

• Add support for reading .bdip Xfit binary dipole fit files in mne.read_dipole() by Eric Larson

• Allow returning vector source estimates from sparse inverse solvers through pick_ori='vector' by Christian Brodbeck

• Add NIRS support to mne.viz.plot_topomap() and mne.viz.plot_ica_components() by Robert Luke

• Add the ability to mne.channels.equalize_channels() to also re-order the channels and also operate on instances of mne.Info, mne.Forward, mne.Covariance and mne.time_frequency.CrossSpectralDensity by Marijn van Vliet

• Allow mne.channels.read_custom_montage to handle fiducial points for BESA spherical (.elp) files by Richard Höchenberger

• Add rank argument to mne.fit_dipole() by Eric Larson

• Add function to convert events to annotations mne.annotations_from_events() by Nicolas Barascud

• Add function to calculate scalp coupling index for fNIRS data mne.preprocessing.nirs.scalp_coupling_index() by Robert Luke

• Add item argument to mne.Epochs.get_data() for faster access to NumPy data arrays compared to mne.Epochs.__getitem__() for frequent access on large mne.Epochs objects, by Eric Larson

• More accurate coordinate system for Easycap montages in mne.channels.make_standard_montage() by Christian Brodbeck

• Add border argument to mne.viz.plot_topomap(). border controls the value of the edge points to which topomap values are extrapolated. border='mean' sets these points value to the average of their neighbours. By Mikołaj Magnuski

• Add function mne.viz.link_brains() to link time properties of multiple brain objects interactively by Guillaume Favelier

• Add function mne.preprocessing.annotate_movement() to annotate periods with head motion and mne.preprocessing.compute_average_dev_head_t() to re-estimate the device to head transform with average head position during segments with acceptable head movement. by Adonay Nunes

• Make selected channels more distinguishable in mne.Epochs.plot_sensors() when using kind='select' by Mikołaj Magnuski

• Allow retrieval of the number of Infomax ICA iterations via the new return_n_iter keyword argument of mne.preprocessing.infomax() by Richard Höchenberger

• Expose the number of ICA iterations during the fitting procedure via the n_iter_ attribute of mne.preprocessing.ICA by Richard Höchenberger

• mne.grand_average() now produces a warning when only a single dataset was passed, instead of raising an error by Richard Höchenberger

• Add nrows and ncols parameters to mne.viz.plot_evoked_topomap(), which allows to create multiline topomap plots for Evoked. By Federico Raimondo

• Improve repr of mne.Info (empty items are not shown anymore) by Clemens Brunner

Bug

• Fix the reduce_rank parameter in mne.beamformer.make_lcmv() and mne.beamformer.make_dics() to reduce the rank of the leadfield first before then reducing the rank of the denominator of the beamformer formula for the inversion by Britta Westner and Eric Larson.

• Fix bug with mne.beamformer.make_lcmv() with data where rank was incorrectly checked by Eric Larson.

• Allow mne.channels.read_dig_hpts() to accept point categories (eg. EEG) to be specified in upper case by Alex Gramfort.

• Fix a bug in mne.MixedSourceEstimate.plot_surface that prevented plotting in latest PySurfer by Christian O’Reilly.

• Fix method='eLORETA' in mne.minimum_norm.apply_inverse(), by Eric Larson:

  1. regularization, which was off by roughly a factor of a square root

  2. units, which are now nAm instead of being arbitrary

  3. explained variance and residuals, which were previously unavailable

  4. force_equal in the case of loose orientations, which previously defaulted to True now defaults to False. Reapplying orientation priors at each iteration of the algorithm sufficiently preserves a smooth transition between loose=0. to loose=1. without compromising localization accuracy in testing.

• Fix bug in method='eLORETA' for mne.minimum_norm.apply_inverse() (and variants) to allow restricting source estimation to a label by Luke Bloy

• Fix bug in mne.compute_covariance() and mne.compute_raw_covariance() where biased normalization (based on degrees of freedom) was used and cov.nfree was not set properly by Eric Larson

• Fix mne.VectorSourceEstimate.normal to account for cortical patch statistics using use_cps=True by Eric Larson

• Fix pick_ori='normal' for mne.minimum_norm.apply_inverse() when the inverse was computed with loose=1. and the forward solution was not in surface orientation, by Eric Larson

• Fix missing xlabel for mne.io.Raw.plot_psd() and similar functions when passing a subset of axes from a figure by Alex Gramfort

• Fix wrong DataFrame index when index=None in methods mne.io.Raw.to_data_frame(), mne.Epochs.to_data_frame(), mne.Evoked.to_data_frame(), and mne.SourceEstimate.to_data_frame(), by Daniel McCloy.

• Fix incorrect scaling of cluster temporal extent in mne.stats.summarize_clusters_stc() by Daniel McCloy.

• Fix mne.time_frequency.read_tfrs() to properly read Info transforms by Eric Larson

• Fix mne.viz.plot_sensors() to always plot in head coordinate frame by Daniel McCloy.

• Fix mne.io.read_raw_cnt() for version 1 of format in the presence of annotations by Alex Gramfort.

• Fix mne.decoding.GeneralizingEstimator and related classes to support multi-class sklearn scorers such as 'roc_auc_ovo' and 'roc_auc_ovo_weighted' by Eric Larson

• Fix mne.io.read_raw_ctf() to set measurement date from CTF ds files by Luke Bloy.

• Fix mne.read_epochs_eeglab() that ignored channel locations by Alex Gramfort.

• Fix mne.io.read_raw_brainvision() when channel names have spaces by Sebastian Major.

• Fix mne.io.read_raw_brainvision() when "Core" is in the data header by Eric Larson

• Fix mne.io.Raw.anonymize() to correctly reset raw.annotations.orig_time by Luke Bloy.

• Fix mne.io.Raw.anonymize() to correctly avoid shifting raw.annotations.onset relative to raw.first_samp by Eric Larson

• Fix mne.io.Raw.set_channel_types() and mne.io.Raw.rename_channels() and related methods to return the instance instead of None by Eric Larson

• mne.Epochs.iter_evoked() now does not return a copy of info when copy=False (default parameter) by Alex Gramfort

• The attribute mne.Annotations.orig_time is now read-only, and is a datetime object (or None) rather than float, by Eric Larson

• Fix bug where MRI distances were not properly initialized in mne coreg by Eric Larson

• Fix xscale='log' in mne.io.Raw.plot_psd() and related functions by Alex Gramfort

• Unify behavior of raw.annotations.append(...) when raw.info['meas_date'] is None to make onsets absolute relative to first_samp as they are when raw.info['meas_date'] is not None; i.e., you might need to do raw.annotations.append(old_time + raw.first_time), by Eric Larson

• The info['meas_date'] entry is now a datetime object (or None) rather than float, by Eric Larson

• Fix mne.time_frequency.AverageTFR.save() and related functions to deal with datetime objects properly by Eric Larson

• Fix date reading before Unix time zero (1970 Jan 1) on Windows by Alex Rockhill.

• Fix bug running subprocesses (e.g., in mne.bem.make_watershed_bem()) in Jupyter notebooks, by Eric Larson

• Fix mne.io.read_raw_artemis123() to add HPI locations to digitization points, by Luke Bloy

• Fix mne.Epochs.shift_time() and mne.Evoked.shift_time() to return the modified Epochs or Evoked instance (instead of None) by Daniel McCloy.

• Fix mne.read_annotations() for text files with zero or one annotations, by Adonay Nunes

• Fix bug in ICA where requesting extended infomax via fit_params={'extended': True} was overridden, by Daniel McCloy.

• Fix bug in mne.Epochs.interpolate_bads() where the origin was not used during MEG or EEG data interpolation by Eric Larson. Old behavior can be achieved using origin=(0., 0., 0.) for EEG and origin=(0., 0., 0.04) for MEG, and the new default is origin='auto', which uses a head-digitization-based fit.

• Fix bug in mne.write_evokeds() where evoked.nave was not saved properly when multiple Evoked instances were written to a single file, by Eric Larson

• Fix bug in mne.preprocessing.mark_flat where acquisition skips were not handled properly, by Eric Larson

• Fix bug in mne.viz.plot_bem() where some sources were not plotted by Jean-Rémi King and Eric Larson

• Fix TAL channel parsing (annotations) for EDF-D files by Clemens Brunner

• Fix bug with mne.viz.plot_dipole_locations() when plotting in head coordinates by Eric Larson

• Fix handling in cases where no components are found in mne.preprocessing.corrmap() by Eric Larson

• Fix handling of repeated events in mne.Epochs by Fahimeh Mamashli and Alex Gramfort

• Fix many bugs with plotting sensors overlaid on a head outline. All plotting is now done in head coordinates and scaled by head_radius, which defaults to 0.095, by Eric Larson

• Fix mne.io.anonymize_info() to allow shifting dates of service and to match anticipated changes in mne-cpp by Luke Bloy

• Fix reading of cardinals in .htps files (identifier are int not strings) by Alex Gramfort

• Fix IO of TFRs when event_id contain a / in one of the keys by Alex Gramfort

• Fix plotting of TFRs with linear y-scale (misaligned frequency bins), by Stefan Repplinger and Eric Larson

• Fix info['sfreq'] when decimating in mne.time_frequency.tfr_multitaper() and mne.time_frequency.tfr_morlet() and make sure an error is raised when exceed Nyquist frequency by Adonay Nunes

• Fix bug in EDF(+) loading, filter values ignored by insufficient regex, by Demetres Kostas

• Fix missing scaling of tolerance parameter in mne.inverse_sparse.tf_mixed_norm() and mne.inverse_sparse.mixed_norm(), by Mathurin Massias

• Fix the automatic scaling of the glyphs in mne.viz.plot_vector_source_estimates() by using 10% of the brain width, by Guillaume Favelier

• Fix handling of missing values (NaNs) in mne.time_frequency.psd_welch by Clemens Brunner

• Fix mne.decoding.LinearModel to support the refitted estimator of GridSearchCV in sklearn by Chun-Hui Li

• Fix mne.viz.plot_topomap() when there are bad channels in the provided mne.Info instance by Eric Larson

• Fix check for subject definition existing in source space in mne.compute_source_morph() and related functions by Eric Larson

• Fix creation of new figures in mne.Epochs.plot_image() and mne.viz.plot_epochs_image() by Eric Larson

• Fix bug in Xdawn where filters where selected along the incorrect axis, by Henrich Kolkhorst

• Fix bug in mne.bem.make_watershed_bem() where some surfaces were saved incorrectly in the working directory by Yu-Han Luo

• Fix support for multiple TAL (annotations) channels in BDF reader by Clemens Brunner

• Fix missing brainmask file type in mne.bem.make_watershed_bem() by changing its default value to ws.mgz by Fu-Te Wong

• Fix bug in mne.channels.make_standard_montage() which would return easycap-M1 even when requesting easycap-M10 by Christian Brodbeck

• Fix the way planar gradiometers are combined in mne.viz.plot_tfr_topomap() and mne.Epochs.plot_psd_topomap() by Geoff Brookshire

• Improve error message when trying to load FieldTrip data from a cell array by Thomas Hartmann

• Fix bug in mne.read_evoked_fieldtrip() causing it to crash when channels were present in the provided Info object but were not present in the data by Thomas Hartmann

• Fix placement of extrapolation points in mne.Evoked.plot_topomap() and related functions when exactly three channels were used by Mikołaj Magnuski.

• Fix bug in reading annotations in read_annotations(), which would not accept “;” character by Adam Li

• Fix bug in mne.Report.parse_folder(), which would not recognize *meg.fif files by Dmitrii Altukhov.

• Include fit_params when saving an ICA instance to disk by Richard Höchenberger

• Update old url link in mne.datasets.eegbci.load_data() to EEGMI_URL = 'https://physionet.org/files/eegmmidb/1.0.0/' by Ramiro Gatti

• Ensure corrected p-values calculated by mne.stats.bonferroni_correction() never exceed the value of 1.0 by Richard Höchenberger

API

• mne.io.pick.get_channel_types is now called mne.io.pick.get_channel_type_constants to better reflect its return values, by Daniel McCloy.

• mne.Evoked.as_type() now returns an instance of mne.EvokedArray by Sophie Herbst and Alex Gramfort

• Bumped minimum requirements to fall 2017 versions by Eric Larson:

  • NumPy 1.13.3

  • SciPy 1.0.0

  • matplotlib 2.1

  • scikit-learn 0.19.1 (optional requirement)

  • pandas 0.21 (optional requirement)

• mne.Epochs.plot() now accepts an event_id parameter (useful in tandem with event_colors for specifying event colors by name) by Daniel McCloy.

• Passing ch_types=None has been deprecated in favor of explicitly passing ch_types='misc' in mne.create_info() by Eric Larson

• New time conversion options for methods mne.io.Raw.to_data_frame(), mne.Epochs.to_data_frame(), mne.Evoked.to_data_frame(), and mne.SourceEstimate.to_data_frame(), by Daniel McCloy.

• mne.Epochs.shift_time() and mne.Evoked.shift_time() now allow shifting times by arbitrary amounts (previously only by integer multiples of the sampling period), by Daniel McCloy and Eric Larson.

• The head_pos argument of mne.Evoked.plot_topomap() and related functions has been deprecated in favor of head_radius, by Eric Larson.

• The layout argument to topomap-related functions such as mne.Evoked.plot_topomap() and mne.viz.plot_tfr_topomap() has been deprecated in favor of channel-position based flattening based on the info and sphere argument, by Eric Larson.

• The APIs of mne.io.Raw.plot_projs_topomap(), mne.Epochs.plot_projs_topomap() and mne.Evoked.plot_projs_topomap() are now more similar to mne.viz.plot_projs_topomap() by Daniel McCloy.

• The function mne.setup_volume_source_space() has a sphere_units argument that defaults to 'mm' in 0.20 but will change to 'm' in 0.21, set it to avoid a warning by Eric Larson.

• mne.viz.plot_projs_topomap() and the related methods mne.io.Raw.plot_projs_topomap(), mne.Epochs.plot_projs_topomap() and mne.Evoked.plot_projs_topomap() now accept parameter vlim to control the colormap, with keyword 'joint' computing the colormap jointly across all projectors of a given channel type, by Daniel McCloy.

• New methods mne.io.Raw.get_channel_types(), mne.Epochs.get_channel_types(), mne.Evoked.get_channel_types() by Daniel McCloy.

• Deprecate mne.minimum_norm.point_spread_function and mne.minimum_norm.cross_talk_function by Alex Gramfort

• Deprecate montage parameter of mne.create_info() in favor of mne.io.Raw.set_montage(), mne.Info.set_montage(), and related functions by Eric Larson

• mne.channels.equalize_channels() no longer operates in-place by default, but instead makes copies of the objects when necessary (see copy parameter) by Marijn van Vliet

• mne.channels.equalize_channels() now uses the first object in the list as a template for channel ordering by Marijn van Vliet

• mne.channels.equalize_channels() now also re-orders the channels to match, in addition to dropping channels that are not shared by all objects by Marijn van Vliet

• mne.io.Raw.pick_channels(), mne.Epochs.pick_channels() and mne.Evoked.pick_channels() now have an ordered parameter to enforce the ordering of the picked channels by Marijn van Vliet

• New methods mne.Forward.pick_channels(), mne.Covariance.pick_channels(), mne.Info.pick_channels, mne.time_frequency.CrossSpectralDensity.pick_channels() by Marijn van Vliet

• New attributes mne.Forward.ch_names and mne.Info.ch_names by Marijn van Vliet

• In Xdawn, the components are stored in the rows of attributes filters_ and patterns_ to be consistent with CSP and ICA by Henrich Kolkhorst

• Drop unit keyword argument from mne.channels.read_custom_montage(), as it was unused by Richard Höchenberger

• Deprecate mne.preprocessing.run_ica, use mne.preprocessing.ICA.detect_artifacts instead by Richard Höchenberger

Authors

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

• Adam Li

• Adonay Nunes

• Alex Rockhill

• Alexandre Gramfort

• Ariel Rokem

• Britta Westner

• Carlos de la Torre

• Christian Brodbeck

• Christian O’Reilly

• Chun-Hui Li

• Clemens Brunner

• Daniel McCloy

• Demetres Kostas

• Denis A. Engemann

• Dmitrii Altukhov

• Dominik Welke

• Eric Larson

• Ezequiel Mikulan

• Fahimeh Mamashli

• Fede Raimondo

• Fu-Te Wong

• Geoff Brookshire

• Guillaume Favelier

• Henrich Kolkhorst

• Joan Massich

• José C. García Alanis

• Jussi Nurminen

• Kaisu Lankinen

• Luke Bloy

• Mainak Jas

• Marijn van Vliet

• Mathurin Massias

• Mikolaj Magnuski

• Mohamed Sherif

• Mohammad Daneshzand

• Nataliia Kozhemiako

• Nicolas Barascud

• Nicolas Legrand

• Olaf Hauk

• Padma Sundaram

• Ramiro Gatti

• Richard Höchenberger

• Robert Luke

• Robert Oostenveld

• Sebastian Major

• Sophie Herbst

• Stefan Appelhoff

• Stefan Repplinger

• Thomas Hartmann

• Victor Férat

• Yu-Han Luo