Skip to main content

InvLCMV

Namespace: INVERSELIB  ·  Library: Inverse Library

Python equivalent

mne.beamformer.make_lcmv / apply_lcmv in MNE-Python.

#include <inv/inv_lcmv.h>

class INVLIB::InvLCMV

Linearly Constrained Minimum Variance (LCMV) beamformer for MEG/EEG source localization.

Typical workflow:

// 1. Compute the spatial filter from forward model + data covariance
InvBeamformer filters = InvLCMV::makeLCMV(info, forward, dataCov, noiseCov, reg,
                                           pickOri, weightNorm, reduceRank, inversion);

// 2. Apply to evoked data -> source time courses
InvSourceEstimate stc = InvLCMV::applyLCMV(evoked, filters);

// 3. Or apply to covariance -> source power map
InvSourceEstimate power = InvLCMV::applyLCMVCov(dataCov, filters);

LCMV beamformer (time-domain).

Static Methods​

makeLCMV(info, forward, dataCov, reg, noiseCov, pickOri, weightNorm, reduceRank, invMethod)​

Compute LCMV beamformer spatial filters.

Parameters:

  • info : const FiffInfo & Measurement info (for channel matching).

  • forward : const MNEForwardSolution & Forward solution containing the leadfield.

  • dataCov : const FiffCov & Data covariance matrix (from recording epoch).

  • reg : double Regularization parameter (default 0.05 = 5% of trace).

  • noiseCov : const FiffCov & Noise covariance for whitening. If empty, identity whitening.

  • pickOri : BeamformerPickOri Orientation selection mode (default: None = keep all).

  • weightNorm : BeamformerWeightNorm Weight normalization (default: UnitNoiseGain).

  • reduceRank : bool Reduce leadfield rank by 1 (default: false).

  • invMethod : BeamformerInversion Inversion method for denominator (default: Matrix).

Returns:

  • InvBeamformer — The computed beamformer containing spatial filter weights.

applyLCMV(evoked, filters)​

Apply LCMV beamformer to evoked data to produce source time courses.

stc = W @ (whitener @ proj @ evoked.data)

Parameters:

  • evoked : const FiffEvoked & Evoked data (averaged, channel x time).

  • filters : const InvBeamformer & Pre-computed LCMV beamformer from makeLCMV().

Returns:

applyLCMVRaw(data, tmin, tstep, filters)​

Apply LCMV beamformer to raw data matrix.

Parameters:

  • data : const Eigen::MatrixXd & Data matrix (n_channels, n_times).

  • tmin : float Start time in seconds.

  • tstep : float Time step in seconds.

  • filters : const InvBeamformer & Pre-computed LCMV beamformer from makeLCMV().

Returns:

applyLCMVCov(dataCov, filters)​

Apply LCMV beamformer to a data covariance to produce a source power map.

power_i = trace(W_i @ Cm @ W_i^T)

Parameters:

  • dataCov : const FiffCov & Data covariance matrix.

  • filters : const InvBeamformer & Pre-computed LCMV beamformer from makeLCMV().

Returns:

  • InvSourceEstimate — Source estimate where data column 0 contains power at each source.

applyLCMVEpochs(epochs, tmin, tstep, filters)​

Apply LCMV beamformer to a list of epochs.

Parameters:

  • epochs : const QList< Eigen::MatrixXd > & List of epoch data matrices (n_channels x n_times each).

  • tmin : float Start time of each epoch (seconds).

  • tstep : float Time step (1/sfreq).

  • filters : const InvBeamformer & Pre-computed LCMV beamformer from makeLCMV().

Returns:

makeLCMVResolutionMatrix(forward, info, dataCov, reg, noiseCov)​

Compute the LCMV resolution matrix: R = W @ G.

The resolution matrix describes the point-spread function of the beamformer. Each row shows how a unit source at one location maps through the forward model and back through the beamformer. Ideally R = I (perfect resolution).

Parameters:

  • forward : const MNEForwardSolution & Forward solution with leadfield.

  • info : const FiffInfo & Measurement info.

  • dataCov : const FiffCov & Data covariance.

  • reg : double Regularization (default 0.05).

  • noiseCov : const FiffCov & Noise covariance for whitening (optional).

Returns:

  • Eigen::MatrixXd — Resolution matrix (n_sources x n_sources).

Authors of this file​