Version 0.15 (2017-10-20)

Changelog

• mne.channels.Layout.plot() and mne.viz.plot_layout() now allows plotting a subset of channels with picks argument by Jaakko Leppakangas

• Add .bvef extension (BrainVision Electrodes File) to mne.channels.read_montage by Jean-Baptiste Schiratti

• Add mne.decoding.cross_val_multiscore() to allow scoring of multiple tasks, typically used with mne.decoding.SlidingEstimator, by Jean-Rémi King

• Add mne.decoding.ReceptiveField module for modeling electrode response to input features by Chris Holdgraf

• Add mne.decoding.TimeDelayingRidge class, used by default by mne.decoding.ReceptiveField, to speed up auto- and cross-correlation computations and enable Laplacian regularization by Ross Maddox and Eric Larson

• Add new mne.datasets.mtrf dataset by Chris Holdgraf

• Add example of time-frequency decoding with CSP by Laura Gwilliams

• Add mne.decoding.SPoC to fit and apply spatial filters based on continuous target variables, by Jean-Rémi King and Alexandre Barachant

• Add Fieldtrip’s electromyogram dataset, by Alexandre Barachant

• Add reject_by_annotation option to mne.preprocessing.find_eog_events() (which is also utilised by mne.preprocessing.create_eog_epochs()) to omit data that is annotated as bad by Jaakko Leppakangas

• Add example for fast screening of event-related dynamics in frequency bands by Denis Engemann

• Add mne.time_frequency.EpochsTFR.save() by Jaakko Leppakangas

• Add butterfly mode (toggled with ‘b’ key) to mne.io.Raw.plot() by Jaakko Leppakangas

• Add axes parameter to plot_topo functions by Jaakko Leppakangas

• Add options to change time windowing in mne.chpi.filter_chpi() by Eric Larson

• mne.channels.Montage.plot, mne.channels.DigMontage.plot(), and mne.viz.plot_montage() now allow plotting channel locations as a topomap by Clemens Brunner

• Add background_color parameter to mne.Evoked.plot_topo() and mne.viz.plot_evoked_topo() and improve axes rendering as done in mne.viz.plot_compare_evokeds() by Alex Gramfort

• Add support for GDF files in mne.io.read_raw_edf() by Nicolas Barascud

• Add mne.io.find_edf_events for getting the events as they are in the EDF/GDF header by Jaakko Leppakangas

• Speed up mne.io.Raw.plot() and mne.Epochs.plot() using (automatic) decimation based on low-passing with decim='auto' parameter by Eric Larson and Jaakko Leppakangas

• Add mne.inverse_sparse.mxne_optim.dgap_l21l1 for computing the duality gap for TF-MxNE as the new stopping criterion by Daniel Strohmeier

• Add option to return a list of Dipole objects in sparse source imaging methods by Daniel Strohmeier

• Add mne.inverse_sparse.make_stc_from_dipoles() to generate stc objects from lists of dipoles by Daniel Strohmeier

• Add mne.channels.find_ch_connectivity that tries to infer the correct connectivity template using channel info. If no template is found, it computes the connectivity matrix using Delaunay triangulation of the 2d projected channel positions by Jaakko Leppakangas

• Add IO support for EGI MFF format by Jaakko Leppakangas and ramonapariciog

• Add option to use matplotlib backend when plotting with mne.viz.plot_source_estimates() by Jaakko Leppakangas

• Add mne.channels.Montage.get_pos2d to get the 2D positions of channels in a montage by Clemens Brunner

• Add MGH 60- and 70-channel standard montages to mne.channels.read_montage by Eric Larson

• Add option for embedding SVG instead of PNG in HTML for mne.Report by Eric Larson

• Add confidence intervals, number of free parameters, and χ² to mne.fit_dipole() and mne.read_dipole() by Eric Larson

• mne.SourceEstimate.data is now writable, writing to it will also update mne.SourceEstimate.times by Marijn van Vliet

• mne.io.Raw.plot() and mne.Epochs.plot() now use anti-aliasing to draw signals by Clemens Brunner

• Allow using saved DigMontage to import digitization to mne.gui.coregistration() by Jaakko Leppakangas

• Add function mne.channels.get_builtin_montages() to list all built-in montages by Clemens Brunner

• mne.decoding.SlidingEstimator and mne.decoding.GeneralizingEstimator now accept **fit_params at fitting by Jean-Rémi King

• Add mne.VectorSourceEstimate class which enables working with both source power and dipole orientations by Marijn van Vliet

• Add option pick_ori='vector' to mne.minimum_norm.apply_inverse() to produce mne.VectorSourceEstimate by Marijn van Vliet

• Add support for numpy.random.RandomState argument to seed in statistical clustering functions and better documentation of exact 1-sample tests by Eric Larson

• Extend mne.viz.plot_epochs_image()/mne.Epochs.plot_image() with regards to grouping by or aggregating over channels. See the new example at examples/visualization/plot_roi_erpimage_by_rt.py by Jona Sassenhagen

• Add bootstrapped confidence intervals to mne.viz.plot_compare_evokeds() by Jona Sassenhagen and Denis Engemann

• Add example on how to plot ERDS maps (also known as ERSP) by Clemens Brunner

• Add support for volume source spaces to spatial_src_connectivity and spatio_temporal_src_connectivity by Alex Gramfort

• Plotting raw data (mne.viz.plot_raw() or mne.io.Raw.plot()) with events now includes event numbers (if there are not more than 50 events on a page) by Clemens Brunner

• Add filtering functions mne.Epochs.filter() and mne.Evoked.filter(), as well as pad argument to mne.io.Raw.filter() by Eric Larson

• Add high frequency somatosensory MEG dataset by Jussi Nurminen

• Add reduced set of labels for HCPMMP-1.0 parcellation in mne.datasets.fetch_hcp_mmp_parcellation() by Eric Larson

• Enable morphing between hemispheres with mne.compute_morph_matrix by Christian Brodbeck

• Add return_residual to mne.minimum_norm.apply_inverse() by Eric Larson

• Add return_drop_log to mne.preprocessing.compute_proj_eog() and mne.preprocessing.compute_proj_ecg() by Eric Larson

• Add time cursor and category/amplitude status message into the single-channel evoked plot by Jussi Nurminen

BUG

• Fixed a bug when creating spherical volumetric grid source spaces in setup_volume_source_space() by improving the minimum-distance computations, which in general will decrease the number of used source space points by Eric Larson

• Fix bug in mne.io.read_raw_brainvision() read .vhdr files with ANSI codepage by Okba Bekhelifi and Alex Gramfort

• Fix unit scaling when reading in EGI digitization files using mne.channels.read_dig_montage by Matt Boggess

• Fix picks default in mne.io.Raw.filter() to include ref_meg channels by default by Eric Larson

• Fix mne.decoding.CSP order of spatial filter in patterns_ by Alexandre Barachant

• mne.concatenate_epochs() now maintains the relative position of events during concatenation by Alexandre Barachant

• Fix bug in script mne make_scalp_surfaces by Denis Engemann (this bug prevented creation of high-resolution meshes when they were absent in the first place.)

• Fix writing of raw files with empty set of annotations by Jaakko Leppakangas

• Fix bug in mne.preprocessing.ICA.plot_properties() where merging gradiometers would fail by Jaakko Leppakangas

• Fix mne.viz.plot_sensors() to maintain proper aspect ratio by Eric Larson

• Fix mne.viz.plot_topomap() to allow 0 contours by Jaakko Leppakangas

• Fix mne.preprocessing.ICA source-picking to increase threshold for rank estimation to 1e-14 by Jesper Duemose Nielsen

• Fix mne.set_bipolar_reference() to support duplicates in anodes by Jean-Baptiste Schiratti and Alex Gramfort

• Fix visuals of mne.viz.plot_evoked() and a bug where ylim changes when using interactive topomap plotting by Jaakko Leppakangas

• Fix mne.Evoked.plot_topomap() when using the mask argument with paired gradiometers by Eric Larson

• Fix bug in mne.Label.fill() where an empty label raised an error, by Eric Larson

• Fix mne.io.read_raw_ctf() to also include the samples in the last block by Jaakko Leppakangas

• Fix mne.preprocessing.ICA.save() to close file before attempting to delete it when write fails by Jesper Duemose Nielsen

• Fix mne.simulation.simulate_evoked() to use nave parameter instead of snr, by Yousra Bekhti

• Fix mne.read_bem_surfaces() for BEM files missing normals by Christian Brodbeck

• Fix mne.transform_surface_to() to actually copy when copy=True by Eric Larson

• Fix mne.io.read_raw_brainvision() to read vectorized data correctly by Jaakko Leppakangas and Phillip Alday

• Fix mne.connectivity.spectral_connectivity so that if n_jobs > 1 it does not ignore last n_epochs % n_jobs epochs by Mikołaj Magnuski

• Fix mne.io.read_raw_edf() to infer sampling rate correctly when reading EDF+ files where tal-channel has a higher sampling frequency by Jaakko Leppakangas

• Fix default value of kind='topomap' in mne.channels.Montage.plot to be consistent with mne.viz.plot_montage() by Clemens Brunner

• Fix bug in to_data_frame where non-consecutive picks would make the function crash by Jaakko Leppakangas

• Fix channel picking and drop in mne.time_frequency.EpochsTFR by Lukáš Hejtmánek

• Fix mne.SourceEstimate.transform() to properly update mne.SourceEstimate.times by Marijn van Vliet

• Fix mne.viz.plot_evoked_joint() to allow custom titles without appending information about the channels by Jaakko Leppakangas

• Fix writing a forward solution after being processed by mne.forward.restrict_forward_to_label() or mne.forward.restrict_forward_to_stc() by Marijn van Vliet

• Fix bug in mne.viz.plot_compare_evokeds() where truncate_yaxis was ignored (default is now False), by Jona Sassenhagen

• Fix bug in mne.viz.plot_evoked() where all xlabels were removed when using spatial_colors=True, by Jesper Duemose Nielsen

• Fix field mapping mne.make_field_map() and MEG bad channel interpolation functions (e.g., mne.Evoked.interpolate_bads()) to choose a better number of components during pseudoinversion when few channels are available, by Eric Larson

• Fix bug in mne.io.read_raw_brainvision(), changed default to read coordinate information if available and added test, by Jesper Duemose Nielsen

• Fix bug in mne.SourceEstimate.to_original_src() where morphing failed if two vertices map to the same target vertex, by Marijn van Vliet

• Fix mne.preprocessing.Xdawn to give verbose error messages about rank deficiency and handle transforming mne.Evoked, by Eric Larson

• Fix bug in DC and Nyquist frequency multitaper PSD computations, e.g. in mne.time_frequency.psd_multitaper, by Eric Larson

• Fix default padding type for mne.Epochs.resample() and mne.Evoked.resample() to be 'edge' by default, by Eric Larson

• Fix mne.inverse_sparse.mixed_norm(), mne.inverse_sparse.tf_mixed_norm() and mne.inverse_sparse.gamma_map() to work with volume source space and sphere head models in MEG by Alex Gramfort and Yousra Bekhti

• Fix mne.Evoked.as_type() channel renaming to append '_v' instead of '_virtual' to channel names to comply with shorter naming (15 char) requirements, by Eric Larson

• Fix treatment of CTF HPI coils as fiducial points in mne.gui.coregistration() by Eric Larson

• Fix resampling of events along with raw in mne.io.Raw() to now take into consideration the value of first_samp by Chris Bailey

• Fix labels of PSD plots in mne.viz.plot_raw_psd() by Alejandro Weinstein

• Fix depth weighting of sparse solvers (mne.inverse_sparse.mixed_norm(), mne.inverse_sparse.tf_mixed_norm() and mne.inverse_sparse.gamma_map()) with free orientation source spaces to improve orientation estimation by Alex Gramfort and Yousra Bekhti

• Fix the threshold in mne.beamformer.rap_music() to properly estimate the rank by Yousra Bekhti

• Fix treatment of vector inverse in mne.minimum_norm.apply_inverse_epochs() by Emily Stephen

• Fix mne.find_events() when passing a list as stim_channel parameter by Alex Gramfort

• Fix parallel processing when computing covariance with shrinkage estimators by Denis Engemann

API

• Removed the mne.datasets.megsim dataset because it was taken down by its host (MRN). Use the mne.simulation functions instead, by Eric Larson

• Add skip_by_annotation to mne.io.Raw.filter() to process data concatenated with e.g. mne.concatenate_raws() separately. This parameter will default to the old behavior (treating all data as a single block) in 0.15 but will change to skip_by_annotation='edge', which will separately filter the concatenated chunks separately, in 0.16. This should help prevent potential problems with filter-induced ringing in concatenated files, by Eric Larson

• ICA channel names have now been reformatted to start from zero, e.g. "ICA000", to match indexing schemes in mne.preprocessing.ICA and related functions, by Stefan Repplinger and Eric Larson

• Add mne.beamformer.make_lcmv() and mne.beamformer.apply_lcmv(), mne.beamformer.apply_lcmv_epochs(), and mne.beamformer.apply_lcmv_raw() to enable the separate computation and application of LCMV beamformer weights by Britta Westner, Alex Gramfort, and Denis Engemann.

• Add weight_norm parameter to enable both unit-noise-gain beamformer and neural activity index (weight normalization) and make whitening optional by allowing noise_cov=None in mne.beamformer.lcmv, mne.beamformer.lcmv_epochs, and mne.beamformer.lcmv_raw, by Britta Westner, Alex Gramfort, and Denis Engemann.

• Add new filtering mode fir_design='firwin' (default in the next 0.16 release) that gets improved attenuation using fewer samples compared to fir_design='firwin2' (default in the current 0.15 release) by Eric Larson

• Make the goodness of fit (GOF) of the dipoles returned by mne.beamformer.rap_music() consistent with the GOF of dipoles returned by mne.fit_dipole() by Alex Gramfort.

• mne.decoding.SlidingEstimator will now replace mne.decoding.TimeDecoding to make it generic and fully compatible with scikit-learn, by Jean-Rémi King and Alex Gramfort

• mne.decoding.GeneralizingEstimator will now replace mne.decoding.GeneralizationAcrossTime to make it generic and fully compatible with scikit-learn, by Jean-Rémi King and Alex Gramfort

• mne.viz.decoding.plot_gat_times, mne.viz.decoding.plot_gat_matrix are now deprecated. Use matplotlib instead as shown in the examples, by Jean-Rémi King and Alex Gramfort

• Add norm_trace parameter to control single-epoch covariance normalization in mne.decoding.CSP, by Jean-Rémi King

• Allow passing a list of channel names as show_names in function mne.viz.plot_sensors() and methods mne.Evoked.plot_sensors(), mne.Epochs.plot_sensors() and mne.io.Raw.plot_sensors() to show only a subset of channel names by Jaakko Leppakangas

• Make function mne.io.eeglab.read_events_eeglab public to allow loading overlapping events from EEGLAB files, by Jona Sassenhagen.

• mne.find_events() mask_type parameter will change from 'not_and' to 'and' in 0.16.

• Instead of raising an error, duplicate channel names in the data file are now appended with a running number by Jaakko Leppakangas

• mne.io.read_raw_edf() has now 'auto' option for stim_channel (default in version 0.16) that automatically detects if EDF annotations or GDF events exist in the header and constructs the stim channel based on these events by Jaakko Leppakangas

• mne.io.Raw.plot_psd() now rejects data annotated bad by default. Turn off with reject_by_annotation=False, by Eric Larson

• mne.set_eeg_reference() and the related methods (e.g., mne.io.Raw.set_eeg_reference()) have a new argument projection, which if set to False directly applies an average reference instead of adding an SSP projector, by Clemens Brunner

• Deprecate plot_trans in favor of mne.viz.plot_alignment() and add bem parameter for plotting conductor model by Jaakko Leppakangas

• mne.beamformer.tf_lcmv now has a raw parameter to accommodate epochs objects that already have data loaded with preload=True, with mne.Epochs.load_data(), or that are read from disk, by Eric Larson

• mne.time_frequency.psd_welch and mne.time_frequency.psd_array_welch() now use a Hamming window (instead of a Hann window) by Clemens Brunner

• picks parameter in mne.beamformer.lcmv, mne.beamformer.lcmv_epochs, mne.beamformer.lcmv_raw, mne.beamformer.tf_lcmv and mne.beamformer.rap_music() is now deprecated and will be removed in 0.16, by Britta Westner, Alex Gramfort, and Denis Engemann.

• The keyword argument frequencies has been deprecated in favor of freqs in various time-frequency functions, e.g. mne.time_frequency.tfr_array_morlet(), by Eric Larson

• Add patterns=False parameter in mne.decoding.ReceptiveField. Turn on to compute inverse model coefficients, by Nicolas Barascud

• The scale, scale_time, and unit parameters have been deprecated in favor of scalings, scalings_time, and units in mne.viz.plot_evoked_topomap() and related functions, by Eric Larson

• loose parameter in inverse solvers has now a default value 'auto' depending if the source space is a surface, volume, or discrete type by Alex Gramfort and Yousra Bekhti

• The behavior of 'mean_flip' label-flipping in mne.extract_label_time_course() and related functions has been changed such that the flip, instead of having arbitrary sign, maximally aligns in the positive direction of the normals of the label, by Eric Larson

• Deprecate force_fixed and surf_ori in mne.read_forward_solution() by Daniel Strohmeier

• mne.convert_forward_solution() has a new argument use_cps, which controls whether information on cortical patch statistics is applied while generating surface-oriented forward solutions with free and fixed orientation by Daniel Strohmeier

• mne.write_forward_solution() writes a forward solution as a forward solution with free orientation in X/Y/Z RAS coordinates if it is derived from a forward solution with free orientation and as a forward solution with fixed orientation in surface-based local coordinates otherwise by Daniel Strohmeier

• loose=None in inverse solvers is deprecated, use explicitly loose=0 for fixed constraint and loose=1.0 for free orientations by Eric Larson

• Zero-channel-value in PSD calculation in mne.viz.plot_raw_psd() has been relaxed from error to warning by Alejandro Weinstein

• Expose “rank” parameter in mne.viz.plot_evoked_white() to correct rank estimates on the spot during visualization by Denis Engemann, Eric Larson, Alex Gramfort.

• Show channel name under mouse cursor on topography plots by Jussi Nurminen

• Return maximum response amplitude from mne.Evoked.get_peak()

Authors

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

• akshay0724

• Alejandro Weinstein

• Alexander Rudiuk

• Alexandre Barachant

• Alexandre Gramfort

• Andrew Dykstra

• Britta Westner

• Chris Bailey

• Chris Holdgraf

• Christian Brodbeck

• Christopher Holdgraf

• Clemens Brunner

• Cristóbal Moënne-Loccoz

• Daniel McCloy

• Daniel Strohmeier

• Denis A. Engemann

• Emily P. Stephen

• Eric Larson

• Fede Raimondo

• Jaakko Leppakangas

• Jean-Baptiste Schiratti

• Jean-Rémi King

• Jesper Duemose Nielsen

• Joan Massich

• Jon Houck

• Jona Sassenhagen

• Jussi Nurminen

• Laetitia Grabot

• Laura Gwilliams

• Luke Bloy

• Lukáš Hejtmánek

• Mainak Jas

• Marijn van Vliet

• Mathurin Massias

• Matt Boggess

• Mikolaj Magnuski

• Nicolas Barascud

• Nicole Proulx

• Phillip Alday

• Ramonapariciog Apariciogarcia

• Robin Tibor Schirrmeister

• Rodrigo Hübner

• 19. 13. Gutstein

• Simon Kern

• Teon Brooks

• Yousra Bekhti