mle-toolbox-eeg-meg

name	mle-toolbox-eeg-meg
description	MLE-Toolbox: Comprehensive open-source MATLAB toolbox for end-to-end EEG/MEG analysis with source localization, connectivity analysis, and ML classifiers. Activation: MLE-Toolbox, EEG analysis, MEG analysis, source localization, brain network analysis, neuroimaging toolbox.

MLE-Toolbox: Comprehensive EEG/MEG Analysis

Open-source MATLAB toolbox integrating full EEG/MEG analysis pipeline with preprocessing, source localization, functional connectivity, and machine learning classification.

Metadata

• Source: arXiv:2604.16463
• Authors: Xiaobo Liu
• Published: 2026-04-08
• Categories: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)

Core Methodology

Design Philosophy

Inspired by established neuroimaging platforms (Brainstorm, FieldTrip, EEGLAB), MLE-Toolbox provides unified, user-friendly GUI for complete EEG/MEG analysis workflow.

Full Analysis Pipeline

Raw Data → Preprocessing → Source Localization → Connectivity → ML Classification

Key Features

1. Preprocessing

Automated Artifact Rejection:

• Independent Component Analysis (ICA)
• Signal-Space Projection (SSP)
• Signal-Space Separation (SSS)

2. Source Localization Methods

Multiple inverse solutions:

• MNE: Minimum Norm Estimation
• dSPM: Dynamic Statistical Parametric Mapping
• sLORETA: Standardized Low-Resolution Brain Electromagnetic Tomography
• Beamforming: Adaptive spatial filtering

3. Parcellation & Visualization

• Multi-atlas parcellation
• Anatomical visualization
• Brain region segmentation

4. Spectral Analysis

• Spectral power analysis
• Frequency-band brain mapping
• Phase-Amplitude Coupling (PAC)

5. Connectivity Analysis

• Graph-theoretic brain network analysis
• Functional connectivity estimation
• Network topology metrics

6. Machine Learning

• Integrated classifiers
• Deep learning integration
• Feature extraction pipelines

Implementation Guide

Prerequisites

• MATLAB (R2018b or later recommended)
• Signal Processing Toolbox
• Statistics and Machine Learning Toolbox
• GPU support optional (for deep learning)

Installation

% Clone or download MLE-Toolbox
% Add to MATLAB path
addpath(genpath('/path/to/mle-toolbox'))
mle_startup

Basic Workflow

1. Data Import

% Import raw EEG/MEG data
data = mle_import('filename.eeg', 'format', 'eeglab');

2. Preprocessing

% Artifact rejection with ICA
[data_clean, ica_weights] = mle_ica_reject(data, 'n_components', 32);

% SSP for environmental noise
[data_clean] = mle_ssp(data_clean, 'n_projectors', 8);

3. Source Localization

% Compute forward model
headmodel = mle_headmodel(subject);

% Minimum norm estimation
sources = mne_solve(data_clean, headmodel, 'method', 'mne');

% Or beamforming
sources = mne_solve(data_clean, headmodel, 'method', 'beamformer');

4. Connectivity Analysis

% Compute connectivity matrix
conn_matrix = mle_connectivity(sources, 'method', 'pli');

% Graph theory metrics
metrics = mle_graph_metrics(conn_matrix, 'metrics', {'clustering', 'path_length'});

5. Machine Learning

% Train classifier
[model, accuracy] = mle_classify(features, labels, 'method', 'svm');

% Cross-validation
[model, cv_accuracy] = mle_classify_cv(features, labels, 'kfold', 10);

Advanced Features

Interoperability

Native compatibility with major platforms:

• Brainstorm: Import/export protocols
• FieldTrip: Data format conversion
• EEGLAB: .set file support
• FreeSurfer: Surface and volume import

Report Generation

One-click academic report generation:

mle_generate_report(analysis_results, 'format', 'pdf', 'template', 'neuroimage');

Interactive Visualization

% Brain surface visualization
mle_visualize_brain(sources, 'surface', 'pial', 'colormap', 'jet');

% Time-frequency plots
mle_plot_tfr(data, 'channels', {'Cz', 'Pz', 'Oz'});

% Connectivity graphs
mle_plot_connectivity(conn_matrix, 'atlas', 'aal');

Applications

Clinical Research

• Epilepsy source localization
• Stroke rehabilitation monitoring
• Sleep disorder analysis

Cognitive Neuroscience

• Working memory studies
• Attention network analysis
• Language processing research

BCI Development

• Motor imagery decoding
• P300 spellers
• Error-related potential detection

Drug Development

• Pharmacological EEG studies
• Biomarker discovery
• Treatment response monitoring

Comparison with Existing Tools

Feature	MLE-Toolbox	Brainstorm	FieldTrip	EEGLAB
GUI	✓ Integrated	✓	✗	✓
Source Loc.	MNE, dSPM, sLORETA, Beamforming	Dipole, MNE	Beamforming	Dipole
Connectivity	✓ Graph-theoretic	✓	✓	Limited
ML Integration	✓ Built-in	✗	✗	Limited
Report Gen	✓ One-click	Manual	Manual	✓
Interoperability	All major tools	Limited	Limited	EEGLAB only

Pitfalls

Common Issues

• Memory: Large datasets may require downsampling or chunking
• Head Models: Accurate source localization requires precise head models
• ICA Interpretation: Requires expertise to identify artifact components

Best Practices

1. Always inspect raw data before preprocessing
2. Use subject-specific head models when possible
3. Validate source localization with known functional landmarks
4. Apply appropriate statistical corrections for multiple comparisons

License

Freely available for non-commercial use.

Related Skills

• eeg-hopfield-emotion-energy
• eeg-structure-guided-diffusion
• functional-connectivity-graph-neural-networks

References

• Liu, X. (2026). MLE-Toolbox: An Open-Source Toolbox for Comprehensive EEG and MEG Data Analysis. arXiv:2604.16463.
• Brainstorm: https://neuroimage.usc.edu/brainstorm
• FieldTrip: https://www.fieldtriptoolbox.org
• EEGLAB: https://sccn.ucsd.edu/eeglab