Version 0.12 (2016-05-10)

Changelog

• Add overlay_times parameter to mne.viz.plot_epochs_image() to be able to display for example reaction times on top of the images, by Alex Gramfort

• Animation for evoked topomap in mne.Evoked.animate_topomap() by Jaakko Leppakangas

• Make mne.channels.find_layout() more robust for KIT systems in the presence of bad or missing channels by Jaakko Leppakangas

• Add raw movement compensation to mne.preprocessing.maxwell_filter() by Eric Larson

• Add mne.Annotations for for annotating segments of raw data by Jaakko Leppakangas

• Add reading of .fif file montages by Eric Larson

• Add system config utility mne.sys_info() by Eric Larson

• Automatic cross-validation and scoring metrics in mne.decoding.GeneralizationAcrossTime, by Jean-Remi King

• mne.decoding.GeneralizationAcrossTime accepts non-deterministic cross-validations, by Jean-Remi King

• Add plotting RMS of gradiometer pairs in mne.viz.plot_evoked_topo() by Jaakko Leppakangas

• Add regularization methods to mne.compute_raw_covariance() by Eric Larson.

• Add command mne show_info to quickly show the measurement info from a .fif file from the terminal by Alex Gramfort.

• Add creating forward operator for dipole object mne.make_forward_dipole() by Chris Bailey

• Add reading and estimation of fixed-position dipole time courses (similar to Elekta xfit) using mne.read_dipole() and mne.fit_dipole() by Eric Larson.

• Accept mne.decoding.GeneralizationAcrossTime’s scorer parameter to be a string that refers to a scikit-learn metric scorer by Asish Panda.

• Add method mne.Epochs.plot_image() calling mne.viz.plot_epochs_image() for better usability by Asish Panda.

• Add mne.io.read_raw_cnt() for reading Neuroscan CNT files by Jaakko Leppakangas

• Add decim parameter to mne.time_frequency.cwt_morlet, by Jean-Remi King

• Add method mne.Epochs.plot_topo_image() by Jaakko Leppakangas

• Add the ability to read events when importing raw EEGLAB files, by Jona Sassenhagen.

• Add function mne.viz.plot_sensors() and methods mne.Epochs.plot_sensors(), mne.io.Raw.plot_sensors() and mne.Evoked.plot_sensors() for plotting sensor positions and mne.viz.plot_layout() and mne.channels.Layout.plot() for plotting layouts by Jaakko Leppakangas

• Add epoch rejection based on annotated segments by Jaakko Leppakangas

• Add option to use new-style MEG channel names in mne.read_selection by Eric Larson

• Add option for proj in mne.EpochsArray by Eric Larson

• Enable the usage of mne.viz.plot_topomap() with an mne.Info instance for location information, by Jona Sassenhagen.

• Add support for electrocorticography (ECoG) channel type by Eric Larson

• Add option for first_samp in mne.make_fixed_length_events() by Jon Houck

• Add ability to auto-scale channel types for mne.viz.plot_raw() and mne.viz.plot_epochs() and corresponding object plotting methods by Chris Holdgraf

BUG

• mne.time_frequency.compute_raw_psd, mne.time_frequency.compute_epochs_psd, mne.time_frequency.psd_multitaper, and mne.time_frequency.psd_welch no longer remove rows/columns of the SSP matrix before applying SSP projectors when picks are provided by Chris Holdgraf.

• mne.Epochs.plot_psd() no longer calls a Welch PSD, and instead uses a Multitaper method which is more appropriate for epochs. Flags for this function are passed to mne.time_frequency.psd_multitaper by Chris Holdgraf

• Time-cropping functions (e.g., mne.Epochs.crop(), mne.Evoked.crop(), mne.io.Raw.crop(), mne.SourceEstimate.crop()) made consistent with behavior of tmin and tmax of mne.Epochs, where nearest sample is kept. For example, for MGH data acquired with sfreq=600.614990234, constructing Epochs(..., tmin=-1, tmax=1) has bounds +/-1.00064103, and now epochs.crop(-1, 1) will also have these bounds (previously they would have been +/-0.99897607). Time cropping functions also no longer use relative tolerances when determining the boundaries. These changes have minor effects on functions that use cropping under the hood, such as mne.compute_covariance() and mne.connectivity.spectral_connectivity. Changes by Jaakko Leppakangas and Eric Larson

• Fix EEG spherical spline interpolation code to account for average reference by Mainak Jas

• MEG projectors are removed after Maxwell filtering by Eric Larson

• Fix mne.decoding.TimeDecoding to allow specifying clf by Jean-Remi King

• Fix bug with units (uV) in ‘Brain Vision Data Exchange Header File Version 1.0’ by Federico Raimondo

• Fix bug where mne.preprocessing.maxwell_filter() destination parameter did not properly set device-to-head transform by Eric Larson

• Fix bug in rank calculation of mne.utils.estimate_rank, mne.io.Raw.estimate_rank, and covariance functions where the tolerance was set to slightly too small a value, new ‘auto’ mode uses values from scipy.linalg.orth by Eric Larson.

• Fix bug when specifying irregular train_times['slices'] in mne.decoding.GeneralizationAcrossTime, by Jean-Remi King

• Fix colorbar range on norm data by Jaakko Leppakangas

• Fix bug in mne.preprocessing.run_ica, which used the ecg_criterion parameter for the EOG criterion instead of eog_criterion by Christian Brodbeck

• Fix normals in CTF data reader by Eric Larson

• Fix bug in mne.io.read_raw_ctf(), when omitting samples at the end by Jaakko Leppakangas

• Fix info['lowpass'] value for downsampled raw data by Eric Larson

• Remove measurement date from mne.Info in mne.io.Raw.anonymize() by Eric Larson

• Fix bug that caused synthetic ecg channel creation even if channel was specified for ECG peak detection in mne.preprocessing.create_ecg_epochs() by Jaakko Leppakangas

• Fix bug with vmin and vmax when None is passed in mne.viz.plot_topo_image_epochs() by Jaakko Leppakangas

• Fix bug with mne.label_sign_flip() (and mne.extract_label_time_course()) by Natalie Klein and Eric Larson

• Add copy parameter in mne.Epochs.apply_baseline() and mne.io.Raw.filter() methods by Jona Sassenhagen and Alex Gramfort

• Fix bug in mne.merge_events() when using replace_events=False by Alex Gramfort

• Fix bug in mne.Evoked type setting in mne.stats.linear_regression_raw() by Eric Larson

• Fix bug in mne.io.edf.RawEDF highpass filter setting to take max highpass to match warning message by Teon Brooks

• Fix bugs with coordinane frame adjustments in mne.viz.plot_trans by Eric Larson

• Fix bug in colormap selection in mne.Evoked.plot_projs_topomap() by Jaakko Leppakangas

• Fix bug in source normal adjustment that occurred when 1) patch information is available (e.g., when distances have been calculated) and 2) points are excluded from the source space (by inner skull distance) by Eric Larson

• Fix bug when merging info that has a field with list of dicts by Jaakko Leppakangas

• The BTi/4D reader now considers user defined channel labels instead of the hard-ware names, however only for channels other than MEG. By Denis Engemann and Alex Gramfort.

• The BTi reader mne.io.read_raw_bti() can now read 2500 system data, by Eric Larson

• Fix bug in mne.compute_raw_covariance() where rejection by non-data channels (e.g. EOG) was not done properly by Eric Larson.

• Change default scoring method of mne.decoding.GeneralizationAcrossTime and mne.decoding.TimeDecoding to estimate the scores within the cross-validation as in scikit-learn as opposed to across all cross-validated y_pred. The method can be changed with the score_mode parameter by Jean-Remi King

• Fix bug in mne.io.Raw.save() where, in rare cases, automatically split files could end up writing an extra empty file that wouldn’t be read properly by Eric Larson

• Fix mne.realtime.StimServer by removing superfluous argument ip used while initializing the object by Mainak Jas.

• Fix removal of projectors in mne.preprocessing.maxwell_filter() in st_only=True mode by Eric Larson

API

• The default picks=None in mne.viz.plot_epochs_image() now only plots the first 5 channels, not all channels, by Jona Sassenhagen

• The mesh_color parameter in mne.viz.plot_dipole_locations() has been removed (use brain_color instead), by Marijn van Vliet

• Deprecated functions mne.time_frequency.compute_raw_psd and mne.time_frequency.compute_epochs_psd, replaced by mne.time_frequency.psd_welch by Chris Holdgraf

• Deprecated function mne.time_frequency.multitaper_psd and replaced by mne.time_frequency.psd_multitaper by Chris Holdgraf

• The y_pred attribute in mne.decoding.GeneralizationAcrossTime and mne.decoding.TimeDecoding is now a numpy array, by Jean-Remi King

• The mne.bem.fit_sphere_to_headshape() function now default to dig_kinds='auto' which will use extra digitization points, falling back to extra plus eeg digitization points if there not enough extra points are available.

• The mne.bem.fit_sphere_to_headshape() now has a units argument that should be set explicitly. This will default to units='mm' in 0.12 for backward compatibility but change to units='m' in 0.13.

• Added default parameters in Epochs class namely event_id=None, tmin=-0.2 and tmax=0.5.

• To unify and extend the behavior of mne.compute_raw_covariance() relative to mne.compute_covariance(), the default parameter tstep=0.2 now discards any epochs at the end of the mne.io.Raw instance that are not the full tstep duration. This will slightly change the computation of mne.compute_raw_covariance(), but should only potentially have a big impact if the mne.io.Raw instance is short relative to tstep and the last, too short (now discarded) epoch contained data inconsistent with the epochs that preceded it.

• The default picks=None in mne.io.Raw.filter() now picks eeg, meg, seeg, and ecog channels, by Jean-Remi King and Eric Larson

• EOG, ECG and EMG channels are now plotted by default (if present in data) when using mne.viz.plot_evoked() by Marijn van Vliet

• Replace pseudoinverse-based solver with much faster Cholesky solver in mne.stats.linear_regression_raw(), by Jona Sassenhagen.

• CTF data reader now reads EEG locations from .pos file as HPI points by Jaakko Leppakangas

• Subselecting channels can now emit a warning if many channels have been subselected from projection vectors. We recommend only computing projection vertors for and applying projectors to channels that will be used in the final analysis. However, after picking a subset of channels, projection vectors can be renormalized with mne.Info.normalize_proj() if necessary to avoid warnings about subselection. Changes by Eric Larson and Alex Gramfort.

• Rename and deprecate mne.Epochs.drop_bad_epochs to mne.Epochs.drop_bad(), and mne.Epochs.drop_epochs to mne.Epochs.drop() by Alex Gramfort.

• The C wrapper mne.do_forward_solution has been deprecated in favor of the native Python version mne.make_forward_solution() by Eric Larson

• The events parameter of mne.EpochsArray() is set by default to chronological time-samples and event values to 1, by Jean-Remi King

Authors

The committer list for this release is the following (preceded by number of commits):

• 348 Eric Larson

• 347 Jaakko Leppakangas

• 157 Alexandre Gramfort

• 139 Jona Sassenhagen

• 67 Jean-Remi King

• 32 Chris Holdgraf

• 31 Denis A. Engemann

• 30 Mainak Jas

• 16 Christopher J. Bailey

• 13 Marijn van Vliet

• 10 Mark Wronkiewicz

• 9 Teon Brooks

• 9 kaichogami

• 8 Clément Moutard

• 5 Camilo Lamus

• 5 mmagnuski

• 4 Christian Brodbeck

• 4 Daniel McCloy

• 4 Yousra Bekhti

• 3 Fede Raimondo

• 1 Jussi Nurminen

• 1 MartinBaBer

• 1 Mikolaj Magnuski

• 1 Natalie Klein

• 1 Niklas Wilming

• 1 Richard Höchenberger

• 1 Sagun Pai

• 1 Sourav Singh

• 1 Tom Dupré la Tour

• 1 jona-sassenhagen@

• 1 kambysese

• 1 pbnsilva

• 1 sviter

• 1 zuxfoucault