Skip to main content

MNE Dipole Fit

MNE Dipole Fit is a GUI/CLI application for sequential equivalent current dipole (ECD) fitting. It localizes focal brain activity by fitting one or more dipoles to MEG/EEG evoked data at each time point, finding the source position, orientation, and amplitude that best explain the measured field pattern.

This application is a C++ port of the original MNE-C mne_dipole_fit tool, enhanced with a 3D visualization of fitted dipoles on the cortical surface.

How It Works​

  1. Load data — evoked FIFF file, noise covariance, BEM model (or sphere model), and optional MRI transformation.
  2. Fit dipoles — at each time point in the specified range, the algorithm performs a non-linear least-squares fit to find the dipole position, orientation, and amplitude that minimize the residual field. An initial grid search over precomputed guess points provides starting estimates.
  3. Visualize — fitted dipoles are displayed on the FreeSurfer cortical surface in a 3D viewer, colour-coded by goodness-of-fit.

Command-Line Options​

MNE Dipole Fit accepts the same command-line options as the original MNE-C tool:

OptionDescription
--meas <file>Evoked data file (FIFF)
--set <n>Data set number in the evoked file (default: 1)
--bad <file>File listing bad channel names
--noise <file>Noise covariance file
--gradnoise <val>Gradiometer noise override (fT/cm)
--magnoise <val>Magnetometer noise override (fT)
--eegnoise <val>EEG noise override (µV)
--reg <val>Regularization factor for noise covariance (default: 0.1 for MEG, 0.1 for EEG)
--gradreg <val>Gradiometer regularization override
--magreg <val>Magnetometer regularization override
--eegreg <val>EEG regularization override
--tmin <ms>Start of fitting interval (ms)
--tmax <ms>End of fitting interval (ms)
--tstep <ms>Time step between fitted dipoles (ms)
--integ <ms>Integration time over response (ms)
--bmin <ms>Baseline start (ms)
--bmax <ms>Baseline end (ms)
--bem <file>BEM model file (FIFF)
--origin <x:y:z>Sphere model origin (mm, MEG head frame)
--eegorigin <x:y:z>EEG sphere model origin
--eegscalpScale EEG electrode positions to the scalp
--mri <file>Head-to-MRI transformation file
--accurateUse the accurate (slower) BEM formulation
--guess <file>Custom guess point file
--guessrad <mm>Radius of the guess point sphere (mm)
--guessrr <mm>Minimum distance between guess points (mm)
--dip <file>Output dipole file
--bdip <file>Output binary dipole file (FIFF)

Example​

mne_dipole_fit \
  --meas audvis-ave.fif \
  --set 1 \
  --noise audvis-cov.fif \
  --bem sample-5120-5120-5120-bem-sol.fif \
  --mri sample-head_mri-trans.fif \
  --tmin 40 --tmax 150 --tstep 5 \
  --dip audvis-dip.txt --bdip audvis-dip.fif

Correspondence to MNE-Python​

The equivalent function in MNE-Python is mne.fit_dipole(), which provides identical sequential dipole fitting with additional options for parallel processing and confidence intervals.

Correspondence to MNE-C​

This application replaces the original mne_dipole_fit from MNE-C v2.7. All command-line options are preserved for backward compatibility. The underlying fitting algorithm (grid-search initialization + simplex optimization) is mathematically identical.

See Also​