Version 0.17 (2018-11-19)

Changelog

• Add new tutorial for mne.Annotations and events by Joan Massich and Alex Gramfort

• Add support for saving mne.Annotations as CSV and TXT files by Joan Massich and Alex Gramfort

• Add mne.Epochs.shift_time() that shifts the time axis of mne.Epochs by Thomas Hartmann

• Add mne.viz.plot_arrowmap() computes arrowmaps using Hosaka-Cohen transformation from magnetometer or gradiometer data, these arrows represents an estimation of the current flow underneath the MEG sensors by Sheraz Khan

• Add mne.io.read_raw_fieldtrip(), mne.read_epochs_fieldtrip() and mne.read_evoked_fieldtrip() to import FieldTrip data. By Thomas Hartmann and Dirk Gütlin.

• Add rank parameter to mne.compute_covariance(), mne.cov.regularize() and related functions to preserve data rank and speed up computation using low-rank computations during regularization by Eric Larson and Denis Engemann

• Add new function mne.read_annotations() that can read annotations in EEGLAB, BrainVision, EDF and Brainstorm formats by Joan Massich and Alex Gramfort.

• Add capability to read and save Epochs containing complex data (e.g. after Hilbert-transform) using mne.Epochs.save() and mne.read_epochs(), by Stefan Repplinger, Eric Larson and Alex Gramfort

• Add optically pumped magnetometer dataset and example by Rasmus Zetter and Eric Larson

• Add origin parameter to mne.io.Raw.time_as_index() to allow times to be relative to this origin by Joan Massich

• Add title argument to mne.SourceEstimate.plot() by Eric Larson

• mne.io.Raw.set_annotations() now changes orig_time to match meas_date and shift self.annotations.onset accordingly. Previous behavior is deprecated and would be removed in 0.18. Work by Joan Massich

• Add mne.compute_source_morph() which creates a mne.SourceMorph object to unify morphing any type of source estimates (surface or volume) from one subject to another for group studies. It is now possible to do group studies when working on the volume with MNE. Work by Tommy Clausner during GSOC 2018 with the help of Alex Gramfort and Eric Larson.

• Add ability to pass threshold for EOG to mne.preprocessing.find_eog_events() and mne.preprocessing.create_eog_epochs() by Peter Molfese

• Add possibility to save mne.VolSourceEstimate and mne.MixedSourceEstimate to HDF5 format (file extension .h5) with mne.VolSourceEstimate.save() and mne.MixedSourceEstimate.save() by Alex Gramfort

• Add replace parameter to mne.io.Raw.add_events() to allow adding events while removing the old ones on the stim channel by Alex Gramfort

• Add ability to pass axes to ts_args and topomap_args of mne.viz.plot_evoked_joint() by Jona Sassenhagen

• Add ability to pass a precomputed forward solution to mne.simulation.simulate_raw() by Eric Larson

• Add ability to read and write beamformers with mne.beamformer.read_beamformer() and mne.beamformer.Beamformer.save by Eric Larson

• Add resting-state source power spectral estimation example sphx_glr_auto_examples_datasets_plot_opm_rest_data.py by Eric Larson, Denis Engemann, and Luke Bloy

• Add mne.channels.make_1020_channel_selections() to group 10/20-named EEG channels by location, by Jona Sassenhagen

• Add helmet for Artemis123 for mne.viz.plot_alignment() by Eric Larson

• Add support for reading MATLAB v7.3+ files in mne.io.read_raw_eeglab() and mne.read_epochs_eeglab() via pymatreader by Steven Gutstein, Eric Larson, and Thomas Hartmann

• Add support for raw PSD plots in mne.Report.parse_folder() via raw_psd argument of mne.Report by Eric Larson

• Add trig_shift_by_type parameter in mne.io.read_raw_brainvision() to allow to specify offsets for arbitrary marker types by Henrich Kolkhorst

• Add progress bar support to mne.decoding.SlidingEstimator and mne.decoding.GeneralizingEstimator by Eric Larson

• Add interactive visualization of volume source estimates using mne.viz.plot_volume_source_estimates() by Mainak Jas

• Add mne.head_to_mri() to convert positions from head coordinates to MRI RAS coordinates, by Joan Massich and Alex Gramfort

• Add improved CTF helmet for mne.viz.plot_alignment() by Eric Larson

• Add handling in mne.combine_evoked() and mne.grand_average() for input with the same channels in different orders, if required, by Jona Sassenhagen

• Add split_naming parameter to the mne.io.Raw.save method to allow for BIDS-compatible raw file name construction by Teon Brooks

• Add origin parameter to mne.Evoked.interpolate_bads() and related methods by Eric Larson

• Add automated MEG helmet shape approximation to mne.viz.plot_alignment() by Eric Larson

• Add capability to save a mne.Report to an HDF5 file to mne.Report.save() by Marijn van Vliet

• Add mne.open_report() to read back a mne.Report object that was saved to an HDF5 file by Marijn van Vliet

• Add multi-taper estimation to mne.minimum_norm.compute_source_psd() by Eric Larson

• Add support for custom, e.g. robust, averaging methods in mne.Epochs.average() by Jona Sassenhagen

• Add support for Neuromag 122 system by Alex Gramfort

• Add function mne.io.read_annotations_brainvision for reading directly Brainvision marker files by Alex Gramfort

• Add mne.Report.remove() method to remove existing figures from a report, by Marijn van Vliet

• Add sign to output of max-power orientation for mne.beamformer.make_dics() by Eric Larson

• Add support for pick_ori='max-power' when weight_norm=None in mne.beamformer.make_lcmv() by Marijn van Vliet

• Add support for weight_norm='nai' for all pick_ori options in mne.beamformer.make_lcmv() by Marijn van Vliet

• Add support for weight_norm='nai' to mne.beamformer.make_dics() by Marijn van Vliet

• Add parameter rank=None to mne.beamformer.make_dics() by Marijn van Vliet

• Add parameter rank='full' to mne.beamformer.make_lcmv(), which can be set to None to auto-compute the rank of the covariance matrix before regularization by Marijn van Vliet

• Handle different time vectors in topography plots using mne.viz.plot_evoked_topo() by Jussi Nurminen

• Speed up mne.inverse_sparse.mixed_norm() if the solver parameter is set to bcd using scipy.linalg.get_blas_funcs() by Quentin Bertrand

Bug

• Fix bug with scaling of data in mne.cov._compute_covariance_auto that was affecting the mne.decoding.SPoC estimator by David Sabbagh

• Fix mne.io.Raw.plot_projs_topomap() by Joan Massich

• Fix bug in mne.minimum_norm.compute_source_psd() where the stc.times output was scaled by 1000, by Eric Larson

• Fix default values for 'diagonal_fixed' estimation method of mne.compute_covariance() to be 0.1 for all channel types, as in mne.cov.regularize() by Eric Larson

• Fix reading edf file annotations by Joan Massich

• Fix bug with reading events from BrainVision files by Stefan Appelhoff

• Fix bug where mne.io.read_raw_eeglab() would warn when the stim channel is populated with an array of zeros by Joan Massich

• Fix 2nd column of events in BrainVision to no longer store duration but rather be contained by raw.annotations by Alex Gramfort

• Fix checking of the correctness of the prepared=True argument in mne.minimum_norm.apply_inverse() and related functions by Eric Larson

• Fix bug of not showing ERD’s in baseline rescaled tfr topomaps if grads are combined by Erkka Heinila

• Fix bug with FIF I/O where strings were written in UTF-8 format instead of Latin-1 by Eric Larson

• Fix bug with reading measurement dates from BrainVision files by Stefan Appelhoff

• Fix bug with mne.fit_dipole where the residual was returned as ndarray instead of mne.Evoked instance, by Eric Larson

• Fix bug with mne flash_bem when flash30 is not used by Eric Larson

• Fix bug with mne.stats.permutation_cluster_test() and mne.stats.spatio_temporal_cluster_test() where threshold=None was not calculated properly for a f-oneway test by Daniel McCloy and Eric Larson

• Fix bug with channel names in mgh70 montage in mne.channels.read_montage by Eric Larson

• Fix duplication of info['hpi_meas'] and info['hpi_results'] by Sara Sommariva

• Fix bug in mne.io.read_raw_edf() when reading large files on Windows by Marcin Koculak

• Fix check in mne.viz.plot_sensors() for invalid channel locations by Eric Larson

• Fix bug in mne.io.read_raw_edf() where GDF files had info['highpass'] and info['lowpass'] set to NaN and info['meas_date'] set incorrectly, by Eric Larson

• Fix bug in mne.preprocessing.ICA.apply() to handle arrays as exclude property by Joan Massich

• Fix bug in method='eLORETA' for mne.minimum_norm.apply_inverse() when using a sphere model and saved inv by Eric Larson

• Fix bug in mne.io.Raw where warnings were emitted when objects were deleted by Eric Larson

• Fix vector data support for mne.VolSourceEstimate by Christian Brodbeck

• Fix bug with IIR filtering axis in mne.filter.filter_data() by Eric Larson

• Fix bug with non-boxcar windows in mne.io.Raw.resample() and mne.filter.resample() by Eric Larson

• Fix bug in mne.minimum_norm.apply_inverse() where applying an MEG-only inverse would raise an error about needing an average EEG reference by Eric Larson

• Fix bug in inst.apply_proj() where an average EEG reference was always added by Eric Larson

• Fix bug in mne.time_frequency.tfr_morlet(), mne.time_frequency.tfr_multitaper(), and mne.time_frequency.tfr_stockwell() where not all data channels were picked by Eric Larson

• Fix bug in mne.preprocessing.ICA.plot_overlay() and mne.make_field_map() for CTF data with compensation by Eric Larson

• Fix bug in mne.create_info() passing int as ch_names on Windows by Eric Larson

• Fix bug in mne.realtime.RtEpochs where events during the buildup of the buffer were not correctly processed when incoming data buffers are smaller than the epochs by Henrich Kolkhorst

• Fix bug in mne.io.read_raw_brainvision() where 1-indexed BrainVision events were not being converted into 0-indexed mne events by Steven Bethard

• Fix bug in mne.viz.plot_snr_estimate() and mne.minimum_norm.estimate_snr() where the inverse rank was not properly utilized (especially affecting SSS’ed MEG data) by Eric Larson

• Fix error when saving stc as nifti image when using volume source space formed by more than one label by Alex Gramfort

• Fix error when interpolating MEG channels with compensation using reference channels (like for CTF data) by Alex Gramfort

• Fix bug in mne.make_sphere_model() where EEG sphere model coefficients were not optimized properly by Eric Larson

• Fix bug in mne.io.read_raw_ctf() to read bad channels and segments from CTF ds files by Luke Bloy

• Fix problem with mne.io.Raw.add_channels() where raw.info['bads'] was replicated by Eric Larson

• Fix bug with mne.Epochs where an error was thrown when resizing data (e.g., during mne.Epochs.drop_bad()) by Eric Larson

• Fix naming of raw.info['buffer_size_sec'] to be raw.buffer_size_sec as it is a writing parameter rather than a measurement parameter by Eric Larson

• Fix EGI-MFF parser not to require dateutil package by Eric Larson

• Fix error when running LCMV on MEG channels with compensation using reference channels (like for CTF data) by Alex Gramfort

• Fix the use of sklearn.model_selection.cross_val_predict() with mne.decoding.SlidingEstimator by Alex Gramfort

• Fix event sample number increase when combining many Epochs objects with mne.concatenate_epochs() with by Jasper van den Bosch

• Fix title of custom slider images to mne.Report by Marijn van Vliet

• Fix missing initialization of self._current in mne.Epochs by Henrich Kolkhorst

• Fix processing of data with bad segments and acquisition skips with new skip_by_annotation parameter in mne.preprocessing.maxwell_filter() by Eric Larson

• Fix symlinking to use relative paths in mne flash_bem and mne watershed_bem by Eric Larson

• Fix error in mne coreg when saving with scaled MRI if fiducials haven’t been saved by Ezequiel Mikulan

• Fix normalization error in mne.beamformer.make_lcmv() when pick_ori='normal', weight_norm='unit_noise_gain' by Marijn van Vliet

• Fix MNE-C installation instructions by buildqa

• Fix computation of max-power orientation in mne.beamformer.make_dics() when pick_ori='max-power', weight_norm='unit_noise_gain' by Marijn van Vliet

API

• Deprecated separate reading of annotations and synthesis of STI014 channels in readers by Joan Massich:

  • Deprecated mne.io.read_annotations_eeglab

  • Deprecated annot and annotmap parameters in read_raw_edf()

  • Deprecated stim_channel parameters in read_raw_edf(), read_raw_brainvision(), and read_raw_eeglab()

  Annotations are now added to raw instances directly upon reading as raw.annotations. They can also be read separately with mne.read_annotations() for EEGLAB, BrainVision, EDF, and Brainstorm formats. Use mne.events_from_annotations(raw.annotations) to convert these to events instead of the old way (using STI014 channel synthesis followed by mne.find_events(raw)).

  In 0.17 (this release)

  Use read_raw_...(stim_channel=False) to disable warnings (and stim channel synthesis), but other arguments for stim_channel will still be supported.

  In 0.18

  The only supported option will be read_raw_...(stim_channel=False), and all stim-channel-synthesis arguments will be removed. At this point, stim_channel should be removed from scripts for future compatibility, but stim_channel=False will still be acceptable for backward compatibility.

  In 0.19

  The stim_channel keyword arguments will be removed from read_raw_... functions.

• Calling :meth:mne.io.pick.pick_info removing channels that are needed by compensation matrices (info['comps']) no longer raises RuntimeException but instead logs an info level message. By Luke Bloy

• mne.Epochs.save() now has the parameter fmt to specify the desired format (precision) saving epoched data, by Stefan Repplinger, Eric Larson and Alex Gramfort

• Deprecated mne.SourceEstimate.morph_precomputed, mne.SourceEstimate.morph, mne.compute_morph_matrix, mne.morph_data_precomputed and mne.morph_data in favor of mne.compute_source_morph(), by Tommy Clausner

• Prepare transition to Python 3. This release will be the last release compatible with Python 2. The next version will be Python 3 only.

• CUDA support now relies on CuPy instead of PyCUDA and scikits-cuda. It can be installed using conda install cupy. By Eric Larson

• Functions requiring a color cycle will now default to Matplotlib rcParams colors, by Stefan Appelhoff

• mne.Evoked.plot_image() has gained the ability to show_names, and if a selection is provided to group_by, axes can now receive a dict, by Jona Sassenhagen

• Calling mne.Epochs.decimate() with decim=1 no longer copies the data by Henrich Kolkhorst

• Removed blocking (waiting for new epochs) in mne.realtime.RtEpochs.get_data() by Henrich Kolkhorst

• Warning messages are now only emitted as warnings.warn_explicit() rather than also being emitted as logging messages (unless a logging file is being used) to avoid duplicate warning messages, by Eric Larson

• Deprecated save_stc_as_volume function in favor of mne.VolSourceEstimate.as_volume() and mne.VolSourceEstimate.save_as_volume() by Alex Gramfort

• src.kind now equals to 'mixed' (and not 'combined') for a mixed source space (made of surfaces and volume grids) by Alex Gramfort

• Deprecation of mne.io.Raw.annotations property in favor of mne.io.Raw.set_annotations() by Joan Massich

• The default value of stop_receive_thread in mne.realtime.RtEpochs.stop has been changed to True by Henrich Kolkhorst

• Using the mne.io.Raw.add_channels() on an instance with memmapped data will now resize the memmap file to append the new channels on Windows and Linux, by Eric Larson

• mne.io.Raw.annotations when missing is set to an empty mne.Annotations rather than None by Joan Massich and Alex Gramfort

• Mismatches in CTF compensation grade are now checked in inverse computation by Eric Larson

Authors

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

• Alexandre Gramfort

• Antoine Gauthier

• Britta Westner

• Christian Brodbeck

• Clemens Brunner

• Daniel McCloy

• David Sabbagh

• Denis A. Engemann

• Eric Larson

• Ezequiel Mikulan

• Henrich Kolkhorst

• Hubert Banville

• Jasper J.F. van den Bosch

• Jen Evans

• Joan Massich

• Johan van der Meer

• Jona Sassenhagen

• Kambiz Tavabi

• Lorenz Esch

• Luke Bloy

• Mainak Jas

• Manu Sutela

• Marcin Koculak

• Marijn van Vliet

• Mikolaj Magnuski

• Peter J. Molfese

• Sam Perry

• Sara Sommariva

• Sergey Antopolskiy

• Sheraz Khan

• Stefan Appelhoff

• Stefan Repplinger

• Steven Bethard

• Teekuningas

• Teon Brooks

• Thomas Hartmann

• Thomas Jochmann

• Tom Dupré la Tour

• Tristan Stenner

• buildqa

• jeythekey