Version 1.12.0.dev157+g2d0c20262 (development) (2026-04-07)

Dependencies

• Updated minimum for:

  • Core dependency scipy >= 1.12

  • Optional dependency pandas >= 2.2

  • Optional dependency scikit-learn >= 1.4

  Changed implemented via CI action created by Thomas Binns. (#13611)

• Updated minimum for:

  • Core dependency scipy >= 1.13

  Changed implemented via CI action created by Thomas Binns. (#13814)

Bugfixes

• Fix bug where outdated seghead.mgz files were reused in mne.bem.make_scalp_surfaces(). Add a new parameter reuse_seghead to control whether to reuse existing seghead.mgz files, by Victor Ferat. (#13024)

• Make interpolate_bads() method flexible (ignore, warn, raise) about how to handle interpolation of channels with invalid positions, by Himanshu Mahor. (#13518)

• Preserve kit_system_id in forward-solution Info, by new contributor Melih Yayli. (#13520)

• Fix bug preventing mne.time_frequency.read_spectrum() from reading saved mne.time_frequency.Spectrum objects created from mne.time_frequency.EpochsSpectrum.average(), by Thomas Binns. (#13521)

• Fix bug where mne.chpi.refit_hpi() did not take gof_limit into account when fitting HPI order, by Eric Larson (#13525)

• Fix bug preventing reading of mne.time_frequency.Spectrum and mne.time_frequency.BaseTFR objects created in MNE<1.8 using the deprecated subject info birthday tuple format, by Thomas Binns. (#13526)

• Fix bug where invalid date formats passed to mne.Info.set_meas_date() were not caught, by Thomas Binns. (#13528)

• Fix bug with mne coreg where interactively altered MRI fiducials were not used when scaling a surrogate subject, by Eric Larson. (#13539)

• Fix bug with mne.viz.plot_evoked() where channels were plotted above axis spines, by Michael Straube. (#13549)

• Fix axis limits in mne.preprocessing.eyetracking.Calibration.plot() to use screen resolution if available, by new contributor Gnaneswar Lopinti. (#13558)

• Fix bug with montage test error message checking, by Eric Larson. (#13579)

• Fix bug where mne.viz.plot_evoked_white() did not accept a single “meg” rank value like those returned from mne.compute_rank(), by Eric Larson. (#13595)

• Fix bug where mne.viz.plot_raw() would access incorrect matplotlib attributes, by new contributor Thomas Caswell. (#13606)

• Fix bug with mne.make_forward_solution() when no MEG channels are present where dev_head_t was errantly set, by Eric Larson. (#13619)

• Fix mne.io.read_raw_snirf() raising TypeError when landmarkLabels is empty or scalar, by Bruno Aristimunha. (#13628)

• Fix bug with mne.datasets.somato.data_path() where the archive couldn’t be extracted due to an absolute path issue, by Eric Larson. (#13630)

• Fix bug in mne.Epochs.plot() where clicking on the scrollbar would not scroll to the last epoch, by Clemens Brunner. (#13654)

• Expose mne.bem namespace, which was inadverdently removed in MNE version 1.5. As an example, users can once again use the mne.bem.make_watershed_bem API entrypoint, by Scott Huberty. (#13669)

• Allow reading alternative nasion label (“Nz”) in SNIRF files when parsing landmarkPos3D, by new contributor Aniket Singh Yadav. (#13672)

• Fix sensor name ordering in plot_topomap() when passing an Info object as pos with names argument, by Famous077. (#13686)

• Fix bug in compute_source_morph() where the spacing attribute was set incorrectly when src_to is provided, by Famous077. (#13687)

• Avoid raising an error when setting montage on a DigMontage of MEG+EEG recordings, when EEG reference positions were set to the placeholder value [1, 0, 0] (i.e., digitization was not performed), by Famous077. (#13700)

• Fix annotation removal logic in matplotlib based figures to correctly handle non-contiguous visible annotations, by Johannes Herforth. (#13703)

• Fix bug in 3D overlay compositing that could produce NaN RGBA values when the resulting alpha is zero, by Pragnya Khandelwal. (#13714)

• Change default-pick ‘data’ for “eyegaze” to false, following existing documentation. by new contributor Benedikt Ehinger. (#13723)

• Support mixed-type concatenation of mne.Annotations and mne.HEDAnnotations, preserving HED strings in extras["HED"], by Bruno Aristimunha. (#13736)

• Improved handling of dropped epochs in mne.viz.plot_ica_properties() and mne.preprocessing.ICA.plot_properties(), allowing plots to be generated even with many bad annotations, by Clemens Brunner. (#13746)

• When interpolating from axial to planar gradiometers, set info fields “chs”, “ch_names”, and “nchan”, according to the interpolated data, reset info fields “device_info”, “helium_info”, “gantry_angle”, “ctf_head_t”, “dev_ctf_t”, “bads”, “projs”, and “comps” to [] or None, and preserve all remaining info fields. by new contributor Christoph Huber-Huber. (#13759)

• Clarified an internal forward-model invariant in mne.forward._lead_dots by asserting that rref is None on the current code path, by Pragnya Khandelwal. (#13764)

• Fix bug in when reading EDF/GDF files where non-positive lowpass values were not handled correctly, now mapping them to the Nyquist frequency to prevent errors during plotting, by new contributor Hansuja Budhiraja. (#13769)

• Clearer error message when no ECG events are found in find_ecg_events(), by new contributor Hansuja Budhiraja. (#13771)

• Avoid some unnecessary computations when n_jobs=None is equivalent to n_jobs=1, by Simon Kern. (#13777)

• Fix clipped annotations and x-axis label in mne.io.Raw.plot(), by Clemens Brunner. (#13787)

New features

• Add mne.HEDAnnotations for attaching HED (Hierarchical Event Descriptors) tags to annotations, by new contributor Kay Robbins, Bruno Aristimunha, and Daniel McCloy. (#13059)

• Add support for mne.io.Raw.interpolate_to() and related methods (and mne.channels.read_meg_canonical_info()) for interpolating MEG data across systems, by new contributor Konstantinos Tsilimparis. (#13196)

• Add support for multi-wavelength NIRS processing to mne.preprocessing.nirs.beer_lambert_law(), mne.preprocessing.nirs.scalp_coupling_index(), and SNIRF reader mne.io.read_raw_snirf(), by new contributor Tamas Fehervari. (#13408)

• Add an optional show_channel_names parameter to mne.viz.plot_alignment() to overlay channel labels at sensor locations in the 3D alignment view, by new contributor Aman Srivastava. (#13570)

• Add mne.io.read_raw_mef() for reading MEF3 files, supporting BIDS iEEG data formats, by Bruno Aristimunha. (#13610)

• Add finer-grained control to mne.io.anonymize_info() and the related methods mne.io.Raw.anonymize(), mne.Epochs.anonymize(), and mne.Evoked.anonymize() by allowing keep_his to accept one or more strings; keep_his=True remains unchanged (retaining "his_id", "sex", and "hand" from anonymization), but individual fields can now also be retained (the default keep_his=False also remains unchanged and still anonymizes all fields), by Clemens Brunner. (#13647)

• Add ICA source timecourse plots to mne.Report.add_ica(), by Aniket Singh Yadav. (#13697)

• Add support for reading BCI2000 .dat files via mne.io.read_raw_bci2k(), and an example examples/io/read_bci2k.py for downloading and visualizing BCI2000 data, by new contributor Hansuja Budhiraja. (#13699)

• Add support for scalar or per-vertex alpha (shape (n_vertices,)) in distributed overlays, including validation/errors for invalid shapes, tests, and a visualization example in examples/visualization/brain.py using mne.viz.Brain.add_data(), by new contributor Pragnya Khandelwal. (#13706) (#13706)

• Improve memory usage and runtime of mne.time_frequency.csd_fourier(), mne.time_frequency.csd_multitaper(), mne.time_frequency.csd_array_fourier(), and mne.time_frequency.csd_array_multitaper() by avoiding unnecessary full-matrix CSD construction, by Pragnya Khandelwal. (#13719)

• Add example showing how to interoperate with R using rpy2, by Aman Srivastava. (#13729)

• Added cmap parameter to Evoked.animate_topomap to allow user-specified colormaps., by new contributor Hansuja Budhiraja. (#13756)

• Allow per-channel color overrides in mne.viz.plot_raw() via channel name keys in the color dict, by new contributor Hansuja Budhiraja. (#13765)

• Added a font_file parameter to mne.viz.Brain.add_text() to support rendering glyphs not available in the default font, by Pragnya Khandelwal. (#13778)

Other changes

• Created a quality control report tutorial Quality control (QC) reports with mne.Report, by new contributor varshaa-1616. (#13532)

• Improved docs for mne.io.read_raw_nirx(), by new contributor Natneal B. (#13541)

• Refactor f-strings in mne/tests/test_import_nesting.py, by new contributor Shruti Bhale. (#13551)

• Remove legacy Python-version compatibility checks in mne.utils.misc._empty_hash now that usedforsecurity=False is always supported, by new contributor Varun Kasyap Pentamaraju (#13566). (#13566)

• Fix deprecation of setting a shape on an array directly in numpy 2.5+, by Mathieu Scheltienne. (#13585)

• Documentation improvement for eSSS in docstring mne.preprocessing.maxwell_filter() and in the MaxFilter example by Simon Kern. (#13591)

• make mne.sys_info() work on Windows systems with Powershell version 7+, by Marijn van Vliet (#13593)

• Examples Using <some-method> section quirk fix in documentation, by new contributor Himanshu Mahor. (#13596)

• Code refactor in mne.preprocessing.eyetracking.set_channel_types_eyetrack(), by Himanshu Mahor. (#13602)

• Improve documentation of return values to clarify that methods typically return the original instance type for chaining purposes, by Aniket Singh Yadav. (#13674)

• Add mne.Epochs.reset_index(), which replaces the legacy mne.Epochs.reset_drop_log_selection() (the new name better reflects the method’s purpose), by Clemens Brunner. (#13675)

• Document the behavior of first_samp after crop() and add reset_first_samp parameter to reset it to 0 after cropping, by new contributor Famous077. (#13685)

• Add a note to mne.Epochs.drop() describing a potential mismatch between the indices expected by drop() and those found in the selection attribute when epochs have been automatically dropped, by new contributor Famous077. (#13688)

• Add optional low-variance (“hat”) regularization to mne.stats.f_oneway() via new sigma and method parameters, by Aniket Singh Yadav. (#13698)

• Extended support for visualising cross-spectral densities in mne.viz.plot_csd() and mne.time_frequency.CrossSpectralDensity.plot() to all types of data channels, by new contributor Aniket Singh Yadav. (#13713)

• Add native save and read functionality for mne.decoding.SSD, mne.decoding.CSP, mne.decoding.SPoC, and mne.decoding.XdawnTransformer objects, by Aniket Singh Yadav. (#13718)

• Remove from __future__ import annotations as it is only needed for Python ≤ 3.9 by Aniket Singh Yadav. (#13749)

• Improved internal functions for inspecting MATLAB v7.3 (HDF5) files in mne.io.read_raw_eeglab(), by Bruno Aristimunha. (#13753)

• Made evoked.plot() instantiate MNELineFigure when it creates its own figure, aligning this path with the ongoing 2D plotting figure-class refactor discussed in #7751, by Pragnya Khandelwal. (#13795)

• Replacing abc.abstractclassmethod() with abc.abstractmethod() in mne.viz.Figure3D, by new contributor Himanshu Mahor. (#13806)