MNE-CPP
MNE-CPP is an open-source, cross-platform C++ framework for real-time and offline processing of MEG, EEG, and related neurophysiological data. It is part of The MNE Project, a family of tools originating from Matti Hämäläinen's MNE-C software at the Martinos Center for Biomedical Imaging.
Related MNE projects:
- MNE-Python — Python reimplementation with extended visualization and analysis
- MNE-MATLAB — MATLAB interface for MNE data structures
- MNE-C — the original C implementation (Manual)
Applications
 | MNE Scan — Real-time acquisition and processing of MEG/EEG data. Plugin-based architecture supporting MEGIN, BabyMEG, BrainAmp, eegosports, gUSBAmp, TMSI, Natus, LSL, and FieldTrip Buffer. In active clinical use at Boston Children's Hospital. |
 | MNE Analyze — Sensor- and source-level analysis GUI for pre-recorded data: raw browsing, filtering, averaging, co-registration, dipole fitting, and source localization. |
 | MNE Browse — Interactive viewer for raw MEG/EEG data in FIFF format with multi-channel navigation, scaling, filtering, and channel selection. |
 | MNE Inspect — Interactive 3D brain viewer for FreeSurfer surfaces, BEM models, source estimates, atlases, sensors, and functional connectivity networks. |
Specialist Applications
| Application | Description |
|---|---|
| MNE Analyze Studio | Agent-oriented analysis workbench with an LLM-driven skill host, neuro kernel, and extension SDK for building composable analysis workflows. (Preview — staging branch) |
| MNE Dipole Fit | Sequential equivalent current dipole fitting for localising focal brain activity. C++ port of the original MNE-C mne_dipole_fit with 3D dipole visualization. |
In addition, MNE-CPP ships a set of command-line tools for BEM model creation, forward/inverse computation, data conversion, anonymization, and real-time streaming — all C++ ports of the original MNE-C utilities.
Libraries
All applications are built on MNE-CPP's modular C++ libraries, which depend on Qt 6 and Eigen 5. Optional dependencies include ONNX Runtime (for neural-network inference in the Ml library) and Skigen (header-only ML primitives built on Eigen). Qt and Eigen are fetched automatically by the repository init scripts; neither is bundled in the source tree. The libraries can be used independently to build custom neuroscience applications. See the Library API documentation for details.
| Library | Description |
|---|---|
| Fiff | FIFF file I/O — raw data, epochs, evoked, covariance, projections, events |
| Mne | Core MNE data structures — source spaces, source estimates, hemispheres, forward solutions |
| Fwd | Forward modelling — BEM and MEG/EEG lead-field computation |
| Inv | Inverse source estimation — MNE, dSPM, sLORETA, eLORETA, LCMV/DICS beamformers, RAP MUSIC, dipole fit, HPI |
| Dsp | Digital signal processing — FIR/IIR filtering, ICA, xDAWN, SSS/tSSS, Welch PSD, Morlet TFR, spectrogram, resampling, bad-channel detection, SPHARA, real-time averaging/filtering |
| Connectivity | Functional connectivity — coherence, coherency, PLV, PLI, WPLI, cross-correlation, and network analysis |
| Disp3D | 3D brain visualization (Metal/Vulkan/D3D/OpenGL via Qt RHI) |
| Mna | MNA graph engine — executable analysis pipelines in .mna/.mnx format with typed ports, operator schemas, and batch/stream execution |
| Ml | Machine learning — ONNX Runtime inference, linear models, feature pipelines, scalers, tensors |
| Sts | Statistics — t-tests, F-tests, cluster permutation testing, covariance estimators, source-level metrics |
License: BSD 3-Clause. Versioning: Semantic Versioning.
Getting Involved
MNE-CPP is a community-driven project. Contributions are welcome — see the contributor guide to get started, or browse the GitHub repository.
Research Projects
MNE-CPP has been developed and extended through several funded research projects:
| Project | Duration | Funding | Description |
|---|---|---|---|
| MNE-CE | 2017–2022 | NIH (1U01EB023820) | Device-independent, standardized software for real-time acquisition, control, and processing of electrophysiological data. |
| OCE | 2018–2021 | DFG / FWF (397686322) | Online neuronal connectivity estimation and neurofeedback with transcranial magnetic stimulation (TMS). Real-time MEG/EEG connectivity methods. |
| OSL | 2013–2015 | DFG (Ba 4858/1-1) | Online MEG source estimation using high-performance GPU computing. |
| AWS Credits | 2018–2019 | AWS | Cloud computing support for MNE-CPP via the AWS Credits for Research Program. |
| Azure Credits | 2016–2018 | Microsoft | Cloud computing support via the Microsoft Azure for Research program. |
Funding Organizations
Supporting Institutions
Contact
For questions and feedback, reach out via the MNE Forum or GitHub Issues. You can also contact the core team directly:
| Name | Affiliation | |
|---|---|---|
| Christoph Dinh | Carl Zeiss AG | christoph.dinh@mne-cpp.org |
| Lorenz Esch | Boston Children's Hospital | lorenz.esch@mne-cpp.org |
| Gabriel Motta | gabrielbenmotta@gmail.com | |
| Juan Garcia-Prieto | MGH | jgarciaprieto@mgh.harvard.edu |
| Matti S. Hämäläinen | Martinos Center / MGH | msh@nmr.mgh.harvard.edu |
| Yoshio Okada | Boston Children's Hospital | yoshio.okada@childrens.harvard.edu |
| John C. Mosher | UTHealth Houston | John.C.Mosher@uth.tmc.edu |
| Jens Haueisen | TU Ilmenau | jens.haueisen@tu-ilmenau.de |
| Daniel Baumgarten | Universität Innsbruck | daniel.baumgarten@uibk.ac.at |
For a full list of contributors see the GitHub contributors page.
Legal Notice
MNE-CPP is a community-driven open-source project with no commercial interest. The source code is released under the BSD 3-Clause License.