Version 0.18 (2019-05-19)

Changelog

• Add event_id='auto' in mne.events_from_annotations() to accommodate Brainvision markers by Jona Sassenhagen, Joan Massich and Eric Larson

• Add example on how to simulate raw data using subject anatomy, by Ivana Kojcic, Eric Larson, Samuel Deslauriers-Gauthier and Kostiantyn Maksymenko

• mne.beamformer.apply_lcmv_cov() returns static source power after supplying a data covariance matrix to the beamformer filter by Britta Westner and Marijn van Vliet

• Add butterfly and order arguments to mne.viz.plot_epochs() and offer separated traces for non-meg data (seeg, eeg, ecog) in butterfly view by Stefan Repplinger and Eric Larson

• mne.Epochs.get_data() now takes a picks parameter by Jona Sassenhagen

• plot_compare_evokeds() will generate topo plots if axes='topo' by Jona Sassenhagen

• mne.viz.iter_topography can yield an additional axis, e.g., for plotting legends by Jona Sassenhagen and Daniel McCloy

• Default plot title reflects channel type when picks is a channel type in plot_compare_evokeds() by Daniel McCloy

• Color scale limits in plot_topo_image_epochs() are now computed separately per channel type in combined mag/grad plots, by Daniel McCloy

• mne.simulation.simulate_stc() now allows for label overlaps by Nathalie Gayraud, and Ivana Kojcic

• Add long_format option to the pandas dataframe exporters, e.g mne.Epochs.to_data_frame() by Denis Engemann

• Add example on how to load standard montage Plotting sensor layouts of EEG systems by Joan Massich

• Add new tutorial on EEG forward operator with a template MRI by Alex Gramfort, and Joan Massich

• Add mne.Epochs.apply_hilbert() and mne.Evoked.apply_hilbert() by Eric Larson

• Add convenience fsaverage subject dataset fetcher / updater mne.datasets.fetch_fsaverage() by Eric Larson

• Add fmin and fmax argument to mne.time_frequency.AverageTFR.crop() and to mne.time_frequency.EpochsTFR.crop() to crop TFR objects along frequency axis by Dirk Gütlin

• Add support to mne.read_annotations() to read CNT formats by Joan Massich

• Add reject parameter to mne.preprocessing.ICA.plot_properties() to visualize rejected epochs by Antoine Gauthier

• Add support for picking channels using channel name and type strings to functions with picks arguments, along with a convenience mne.io.Raw.pick(), mne.Epochs.pick(), and mne.Evoked.pick() method, by Eric Larson

• Add new tutorial on Sleep stage classification from polysomnography (PSG) data by Alex Gramfort, Stanislas Chambon and Joan Massich

• Add data fetchers for polysomnography (PSG) recordings from Physionet (mne.datasets.sleep_physionet.age.fetch_data() and mne.datasets.sleep_physionet.temazepam.fetch_data()) by Alex Gramfort and Joan Massich

• Add envelope correlation code in mne.connectivity.envelope_correlation by Denis Engemann, Sheraz Khan, and Eric Larson

• Add option to toggle all projectors in mne.io.Raw.plot() and related functions by Eric Larson

• Add support for indexing, slicing, and iterating mne.Annotations by Joan Massich

• mne.io.Raw.plot() now uses the lesser of n_channels and raw.ch_names, by Joan Massich

• Add support for FIR filtering in mne.io.Raw.plot() and mne browse_raw by passing filtorder=0 or --filtorder 0, respectively, by Eric Larson

• Add chunk_duration parameter to mne.events_from_annotations() to allow multiple events from a single annotation by Joan Massich

• Add mne.simulation.SourceSimulator class to simplify simulating SourceEstimates, by Samuel Deslauriers-Gauthier, Kostiantyn Maksymenko, Nathalie Gayraud, Ivana Kojcic, Alex Gramfort, and Eric Larson

• mne.io.read_raw_edf() now detects analog stim channels labeled 'STATUS' and sets them as stim channel. mne.io.read_raw_edf() no longer synthesize TAL annotations into stim channel but stores them in raw.annotations when reading by Joan Massich

• Add mne.simulation.add_noise for ad-hoc noise addition to io.Raw, Epochs, and Evoked instances, by Eric Larson

• Add drop_refs=True parameter to set_bipolar_reference() to drop/keep anode and cathode channels after applying the reference by Cristóbal Moënne-Loccoz.

• Add processing of reference MEG channels to mne.preprocessing.ICA by Jevri Hanna

• Add use of scipy.signal.windows.dpss() for faster multitaper window computations in PSD functions by Eric Larson

• Add mne.morph_labels() to facilitate morphing label sets obtained from parcellations, by Eric Larson

• Add mne.labels_to_stc() to facilitate working with label data, by Eric Larson

• Add mne.label.select_sources() to simplify the selection of sources within a label, by Samuel Deslauriers-Gauthier

• Add support for using mne.Info in mne.simulation.simulate_raw() instead of mne.io.Raw by Eric Larson

• Add support for passing an iterable and stim channel values using stc parameter of mne.simulation.simulate_raw() by Eric Larson

• Add overlap argument to mne.make_fixed_length_events() by Eric Larson

• Add approximate distance-based spacing source space decimation algorithm to mne.setup_source_space() by Eric Larson

• Add 448-labels subdivided aparc cortical parcellation by Denis Engemann and Sheraz Khan

• Add option to improve rendering in mne coreg for modern graphics cards by Eric Larson

• Add mne.preprocessing.mark_flat to automate marking of flat channels and segments of raw data by Eric Larson

• Add support for CUDA-based correlation computations and progress bars in mne.decoding.ReceptiveField by Eric Larson

• Add support for file-like objects in mne.io.read_raw_fif() as long as preloading is used by Eric Larson

• Add keyboard shortcuts to nativate volume source estimates in time using (shift+)left/right arrow keys by Mainak Jas

• Add option to SSP preprocessing functions (e.g., mne.preprocessing.compute_proj_eog() and mne.compute_proj_epochs()) to process MEG channels jointly with meg='combined' by Eric Larson

• Add Epoch selection and metadata functionality to mne.time_frequency.EpochsTFR using new mixin class by Keith Doelling

• Add reject_by_annotation argument to mne.preprocessing.find_ecg_events() by Eric Larson

• Add pca argument to return the rank-reduced whitener in mne.cov.compute_whitener() by Eric Larson

• Add extrapolate argument to mne.viz.plot_topomap() for better control of extrapolation points placement by Mikołaj Magnuski

• Add channel_wise argument to mne.io.Raw.apply_function() to allow applying a function on multiple channels at once by Hubert Banville

• Add option copy='auto' to control data copying in mne.io.RawArray by Eric Larson

• The mri parameter in mne.setup_volume_source_space() is now automatically set to T1.mgz if subject is provided. This allows to get a mne.SourceSpaces of kind volume more automatically. By Alex Gramfort

• Add better __repr__ for constants, and info['dig'] entries via DigPoint by Eric Larson

• Allow string argument in mne.io.Raw.drop_channels() to remove a single channel by Clemens Brunner

• Add additional depth weighting options for inverse solvers (e.g., mne.inverse_sparse.gamma_map() and mne.inverse_sparse.mixed_norm()) by Eric Larson

• Add depth weighting to LCMV beamformers via depth argument in mne.beamformer.make_lcmv() by Eric Larson

• Allow toggling of DC removal in mne.io.Raw.plot() by pressing the ‘d’ key by Clemens Brunner

• Improved clicking in mne.io.Raw.plot() (left click on trace toggles bad, left click on background sets green line, right click anywhere removes green line) by Clemens Brunner

• Add mne.realtime.LSLClient for realtime data acquisition with LSL streams of data by Teon Brooks and Mainak Jas

• Add partial support for PyVista as a 3D backend using mne.viz.use_3d_backend() by Guillaume Favelier

• Add option ids = None in mne.event.shift_time_events() for considering all events by Nikolas Chalas and Joan Massich

• Add mne.realtime.MockLSLStream to simulate an LSL stream for testing and examples by Teon Brooks

• Add support for file-like objects in mne.read_epochs() as long as preloading is used by Paul Roujansky

Bug

• Fix annotations in split fif files mne.io.read_raw_fif() by Joan Massich

• Fix mne.Epochs.plot() with scalings='auto' to properly compute channel-wise scalings by Stefan Repplinger

• Fix mne.gui.coregistration() and mne coreg crashing with segmentation fault when switching subjects by Eric Larson

• Fix mne.io.read_raw_brainvision() to accommodate vmrk files which do not have any annotations by Alexander Kovrig

• Fix mne.io.Raw.plot() and mne.Epochs.plot() to auto-scale misc channel types by default by Eric Larson

• Fix filtering functions (e.g., mne.io.Raw.filter()) to properly take into account the two elements in n_pad parameter by Bruno Nicenboim

• Fix feature_names parameter change after fitting in mne.decoding.ReceptiveField by Jean-Remi King

• Fix index error in mne.io.read_raw_cnt() when creating stim_channel manually by Joan Massich

• Fix bug with weight_norm='unit-gain' in mne.beamformer.make_lcmv() and mne.beamformer.make_dics() by Britta Westner

• Fix 32bits annotations in mne.io.read_raw_cnt() by Joan Massich

• Fix mne.events_from_annotations() to ignore 'BAD_' and 'EDGE_' annotations by default using a new default regexp by Eric Larson

• Fix bug in mne.preprocessing.mark_flat where raw.first_samp was not taken into account by kalenkovich

• Fix date parsing in mne.io.read_raw_cnt() by Joan Massich

• Fix topological checks and error messages for BEM surfaces in mne.make_bem_model() by Eric Larson

• Fix default HTML language of mne.Report to be "en-us" instead of "fr" and allow setting via report.lang property by Eric Larson

• Fix bug where loading epochs with preload=True and subsequently using mne.Epochs.drop_bad() with new reject or flat entries leads to improper data (and epochs.selection) since v0.16.0 by Eric Larson. If your code uses Epochs(..., preload=True).drop_bad(reject=..., flat=...), we recommend regenerating these data.

• Fix mne flash_bem to properly utilize flash30 images when conversion from DICOM images is used, and to properly deal with non-standard acquisition affines by Eric Larson

• Fix mne.io.Raw.set_annotations() with annotations=None to create an empty annotations object with orig_time that matches the mne.io.Raw instance by Eric Larson

• Fix mne.io.read_raw_edf() returning all the annotations with the same name in GDF files by Joan Massich

• Fix boundaries during plotting of raw data with mne.io.Raw.plot() and mne browse_raw on scaled displays (e.g., macOS with HiDPI/Retina screens) by Clemens Brunner

• Fix bug where filtering was not performed with lowpass or highpass in mne.io.Raw.plot() and mne browse_raw by Eric Larson

• Fix mne.simulation.simulate_evoked() that was failing to simulate the noise with heterogeneous sensor types due to poor conditioning of the noise covariance and make sure the projections from the noise covariance are taken into account Alex Gramfort

• Fix checking of data dimensionality in mne.SourceEstimate and related constructors by Eric Larson

• Fix mne.io.Raw.append() annotations miss-alignment by Joan Massich

• Fix hash bug in the mne.io.edf module when installing on Windows by Eric Larson

• Fix mne.io.read_raw_edf() reading duplicate channel names by Larry Eisenman

• Fix set_bipolar_reference() in the case of generating all bipolar combinations and also in the case of repeated channels in both lists (anode and cathode) by Cristóbal Moënne-Loccoz

• Fix missing code for computing the median when method='median' in mne.Epochs.average() by Cristóbal Moënne-Loccoz

• Fix CTF helmet plotting in mne.viz.plot_evoked_field() by Eric Larson

• Fix saving of rejection parameters in mne.Epochs.save() by Eric Larson

• Fix orientations returned by mne.dipole.get_phantom_dipoles() (half were flipped 180 degrees) by Eric Larson

• Fix bug in mne.viz.plot_compare_evokeds() when evoked.times[0] >= 0 would cause a problem with vlines='auto' mode by Eric Larson

• Fix path bugs in mne.bem.make_flash_bem() and mne flash_bem by Eric Larson

• Fix mne.time_frequency.AverageTFR.plot_joint() mishandling bad channels, by David Haslacher and Jona Sassenhagen

• Fix mne.beamformer.make_lcmv() failing when full rank data is used (i.e., when no projection is done) with reg=0., by Eric Larson

• Fix issue with bad channels ignored in mne.beamformer.make_lcmv() and mne.beamformer.make_dics() by Alex Gramfort

• Fix mne.compute_proj_raw() when duration != None not to apply existing proj and to avoid using duplicate raw data samples by Eric Larson

• Fix reject_by_annotation not being passed internally by mne.preprocessing.create_ecg_epochs() and mne clean_eog_ecg to mne.preprocessing.find_ecg_events() by Eric Larson

• Fix mne.io.read_raw_edf() failing when EDF header fields (such as patient name) contained special characters, by Clemens Brunner

• Fix mne.io.read_raw_eeglab() incorrectly parsing event durations by Clemens Brunner

• Fix mne.io.read_raw_egi() when cropping non-preloaded EGI MFF data by Alex Gramfort

• Fix mne.io.Raw.interpolate_bads() for interpolating CTF MEG channels when reference sensors are present by jeythekey

• Fix a bug in mne.io.Raw.resample(), where resampling events could result in indices > n_times-1, by jeythekey

• Fix mne.preprocessing.ICA.score_sources() to use the sfreq of the raw data to filter rather than the sfreq when the ICA was fit, by jeythekey

• Fix a bug in mne.preprocessing.ICA, where manually setting the attribute ICA.exclude to an np.array resulted in the removal of random components when later also providing the exclude argument to any apply...-method, by jeythekey

• Ascending changed to descending sorting of scores for integer ..._criterion arguments in mne.preprocessing.ICA.detect_artifacts and mne.preprocessing.run_ica, as it used to be documented; the docstring in these functions was corrected for float ..._criterion arguments, by jeythekey

API

• Deprecate cov, iir_params, blink, ecg, chpi, random_state and support for mne.io.Raw instance inputs in mne.simulation.simulate_raw(); use mne.simulation.add_noise(), mne.simulation.add_ecg(), mne.simulation.add_eog(), and mne.simulation.add_chpi() by Eric Larson

• Add overwrite parameter in mne.Epochs.save() by Katarina Slama

• Add stim_channel parameter in mne.io.read_raw_cnt() to toggle stim channel synthesis by Joan Massich

• Python 2 is no longer supported; MNE-Python now requires Python 3.5+, by Eric Larson

• A new class mne.VolVectorSourceEstimate is returned by mne.minimum_norm.apply_inverse() (and related functions) when a volume source space and pick_ori='vector' is used, by Eric Larson

• Converting a forward solution with a volume source space to fixed orientation using mne.convert_forward_solution() now raises an error, by Eric Larson

• raw.estimate_rank has been deprecated and will be removed in 0.19 in favor of mne.compute_rank() by Eric Larson

• Annotations are now kept sorted (by onset time) during instantiation and append() operations, by Eric Larson

• Deprecate mne.io.find_edf_events by Joan Massich

• Deprecate limit_depth_chs in mne.minimum_norm.make_inverse_operator() in favor of depth=dict(limit_depth_chs=...) by Eric Larson

• Reading BDF and GDF files with mne.io.read_raw_edf() is deprecated and replaced by mne.io.read_raw_bdf() and mne.io.read_raw_gdf(), by Clemens Brunner

• mne.forward.compute_depth_prior() has been reworked to operate directly on Forward instance as forward rather than a representation scattered across the parameters G, is_fixed_ori, patch_info, by Eric Larson

• Deprecate method='extended-infomax' in mne.preprocessing.ICA; Extended Infomax can now be computed with method='infomax' and fit_params=dict(extended=True) by Clemens Brunner

• Fix support for supplying extrapolate via ica.plot_properties(..., topomap_args=dict(extrapolate=...)) by Sebastian Castano

• The peak finder that was formerly accessible via from mne.preprocessing.peak_finder import peak_finder should now be imported directly from the enclosing namespace as from mne.preprocessing import peak_finder by Eric Larson

• Deprecate mne.realtime module to make a standalone module mne-realtime that will live outside of this package by Teon Brooks