Data Conversion

This chapter describes the data conversion utilities included with MNE software for importing data from various MEG/EEG systems and converting between different file and coordinate formats.

Importing MEG/EEG System Data

4-D Neuroimaging (Magnes)

The 4-D Magnes software can export data in FIFF format. However, the exported files lack compensation channel information. The complete conversion process involves:

1. Export the raw data FIFF file from the Magnes system
2. If the primary sensors are gradiometers (not magnetometers), run mne_fix_mag_coil_types --magnes to correct channel information
3. Export a text file containing the Magnes compensation sensor data
4. Create a text file containing the compensation channel weights
5. Run mne_insert_4D_comp to merge compensation channel data with the original FIFF file
6. Run mne_create_comp_data to create a FIFF file with the compensation weight matrix
7. Run mne_add_to_meas_info to combine the files into a complete FIFF file

note

Including compensation channel data is recommended but not mandatory. If the data are already compensated in the Magnes system, there will be a small error in the forward calculations.

CTF Data

Use mne_ctf2fiff to convert CTF ds directories to FIFF format. Key options:

Option	Description
--ds <directory>	Read data from this CTF ds directory
--fif <filename>	Output FIFF file name
--omit <filename>	File listing channels to omit from conversion
--evoked	Produce an evoked-response file instead of raw data
--infoonly	Write only measurement info, no data

During conversion, the utility reads .res4 (header info), .hc (HPI coil locations), .meg4 (MEG data), and optionally .eeg (EEG electrode locations) from the ds directory.

important

The conversion includes a transformation from the CTF head coordinate system convention to the Neuromag convention. Epoch mode files with time gaps between epochs are not recommended as they produce discontinuities in the output.

CTF Polhemus data can be converted with mne_ctf_dig2fiff, which reads digitization data in CTF text format and outputs FIFF and hpts formats. Use --numfids if fiducial locations are numbered (1 = nasion, 2 = left auricular, 3 = right auricular) rather than labeled.

Software Gradient Compensation

CTF and 4-D systems use software gradient compensation to suppress external disturbances using reference channel data. The compensation matrices relate uncompensated data $\mathbf{x}^{(0)}$ to compensated data $\mathbf{x}^{(k)}$:

$$\mathbf{x}^{(k)} = \mathbf{M}^{(k)} \mathbf{x}^{(0)}$$

where $\mathbf{M}^{(k)}$ is a composite compensation matrix containing the $n_1 \times n_2$ compensation data matrix $\mathbf{C}^{(k)}$ for grade $k$.

Use mne_compensate_data to change the compensation grade in evoked-response files. For raw data, use mne_process_raw with the desired compensation grade.

KIT MEG System

Use mne_kit2fiff to import 160-channel KIT MEG system data. Required input files:

1. Polhemus data file (.elp) — fiducial and HPI coil locations
2. HPI coil device locations — HPI coil positions in MEG device coordinates
3. Sensor data file (.sns) — sensor locations and orientations

Optional files include a head shape data file (.hsp) and raw data in text format.

The digital trigger channel (STI 014) can be synthesized from analog trigger channels using --stim, with an adjustable threshold (--stimthresh, default 1.0).

EDF, EDF+, and BDF Formats

Use mne_edf2fiff to convert European Data Format (EDF/EDF+, 16-bit) and BioSemi BDF (24-bit) files. Key considerations:

• Electrode locations must be provided separately via --hpts or --elp files (fiducial locations are mandatory)
• EDF+ annotation channels can be converted to trigger channel (STI 014) using --annotmap
• BDF trigger data: use --digtrigmask 0xff to select the significant bytes

Post-conversion steps:

1. Set correct channel types (EOG, EMG) with mne_rename_channels
2. Create topographical layout with mne_make_eeg_layout
3. For BDF files, specify trigger channel name "Status" via --digtrig

warning

BDF 3-byte samples are converted to 4-byte integers. The maximum convertible BDF file size is approximately 1.5 GB. Channel-dependent sampling rates are not supported.

BrainVision EEG Data

Use mne_brain_vision2fiff to import BrainVision data:

Option	Description
--header <name>	BrainVision header file (.vhdr)
--dig <name>	FIFF file with digitization data
--orignames	Keep original EEG channel labels
--eximia	Interpret as eXimia data (threshold first 3 channels as triggers)
--split <size/MB>	Split output at specified file size

The mne_eximia2fiff script provides a convenient wrapper for converting Nexstim eXimia EEG data.

Tufts University Format

The mne_tufts2fiff utility imports Tufts University EEG data from three files:

• Raw data file (.raw)
• Calibration raw data file (c.raw) — needed for physical unit scaling
• Electrode location file (.elp) — fiducial locations are mandatory

Converting Digitization Data

The mne_convert_dig_data utility converts Polhemus digitization data between FIFF, hpts, and elp formats.

The hpts Format

The hpts text format uses lines of the form:

<category> <identifier> <x/mm> <y/mm> <z/mm>

Category	Description
cardinal (or fiducial)	Fiducial landmarks (1 = left auricular, 2 = nasion, 3 = right auricular)
hpi	Head-position indicator coil locations
eeg	EEG electrode locations (identifier 0 = reference electrode)
extra	Additional head surface points

Coordinates are typically in the MEG head coordinate system in millimeters.

Converting Volumetric Data

The mne_volume_data2mri utility converts source estimate data (.w or .stc files) from a volumetric source space into FreeSurfer MRI overlay format (.mgh or .mgz). The volumetric source space must have been created with the --mri option in mne_volume_source_space.

Converting Surface Data

The mne_convert_surface utility converts surface data between multiple formats:

Input Formats	Output Formats
FIFF (--fif)	FIFF (--fifout)
FreeSurfer (--surf)	FreeSurfer (--surfout)
Text triangle file (--tri)	ASCII triangle (--triout)
BrainSuite DFS (--dfs)	ASCII points (--pntout)
	SMF (--smfout)

For DFS files, use --mghmri or --fifmri to specify the MRI volume for coordinate transformation. The --flip option handles left-right coordinate system differences.

Converting MRI Data to FIFF

The mne_make_cor_set utility creates FIFF format MRI description files from FreeSurfer MRI volumes:

# From COR format directory
mne_make_cor_set --dir <directory> --out <output.fif>

# From mgh/mgz format (includes pixel data and transformations)
mne_make_cor_set --mgh <volume.mgz> --out <output.fif>

With --mgh input, the output includes the $T_3$, $T_4$ (Talairach), $T^-$, and $T^+$ coordinate transformations.

Collecting Coordinate Transformations

The mne_collect_transforms utility gathers coordinate transformations from various sources into a single FIFF file. This is useful for assembling all the transformations needed for the analysis pipeline:

Option	Provides
--meas <name>	Head-to-device transformation $T_1$
--mri <name>	Head-to-MRI transformation $T_2$, plus $T_3$, $T_4$, $T^-$, $T^+$
--mgh <name>	MRI volume geometry and Talairach transformation

See Also

• FIFF Format — Details on the FIFF file format
• Forward Solution — Coordinate systems and transformations
• Workflow — The MEG/EEG processing workflow
• Tools Overview — Complete list of command-line tools