Data analysis PC (DANA)

A data analysis PC, called DANA, is available in the MEG facility. DANA can be accessed on the account meguser and has both MEGIN’s analysis software and MNE-Python.

Specifications

The analysis PC run on an intel Xeon W-2245 (8 cores, 16 threads) with 32 GB of RAM and an Nvidia Quadro P2200. Most analysis pipeline require one or multiple copies of a dataset to be loaded in RAM. 32 GB is sufficient for subject-level analysis and most group-level analysis, with an additional 20 GB of available swap if needed.

MEGIN’s software

MaxWell filter

Signal-Space Separation (SSS)[1][2] is a technique based on the physics of electromagnetics fields. SSS separates the measured signal into components attributable to sources inside the measurement volume of the sensor array (the internal components), and components attributable to sources outside the measurement volume (the external components). The internal and external components are linearly independent, so it is possible to simply discard the external components to reduce environmental noise.

Maxwell filtering is a related procedure that omits the higher-order components of the internal subspace, which are dominated by sensor noise. Typically, Maxwell filtering and SSS are performed together.

Like Signal Space Projector[3], SSS is a form of projection. Whereas SSP empirically determines a noise subspace based on data (empty-room recordings, EOG or ECG activity, etc) and projects the measurements onto a subspace orthogonal to the noise, SSS mathematically constructs the external and internal subspaces from spherical harmonics and reconstructs the sensor signals using only the internal subspace (i.e., does an oblique projection).

Introduction taken from MNE’s background on SSS and Maxwell filtering.

Tip

Instead of MEGIN’s old MaxWell software available on the DANA, the enhanced version available in MNE-Python is recommended. See mne.preprocessing.maxwell_filter(). The fine calibration file and the cross-talk correction are available in the sample dataset ( see meg_wiki.datasets.sample.data_path()).

FIFF format

The FIFF format is MEGIN’s proprietary format for MEG data. It is a binary tag based format. The tags define which information is stored in the file. Thus, the FIFF format is not restricted to storing the continuous MEG data. For instance, the Signal Space Projector[3] are stored both in the raw data file and in a separate FIFF file. In both case, the information related to the SSP is stored in the same FIFF tag.

Tip

MNE-Python has command-line tools which can be accessed with:

$ mne <command> <options>

Entering the command $ mne will return a list of available commands.

The commands mne what and mne show_fiff can be used to inspect the content of a FIFF file. For instance, to inspect the content of the file ssp_68_200123_proj.fif:

$ mne what ssp_68_200123_proj.fif
proj

$ mne show_fiff ssp_68_200123_proj.fif
999  = FIFFB_ROOT
    100  = FIFF_FILE_ID (20b ids)  = {'version': 65540, 'machid': a ... dict len=4
    101  = FIFF_DIR_POINTER (4b >i4)  = [-1]
    106  = FIFF_FREE_LIST (4b >i4)  = [-1]
    108  = FIFF_NOP (0b nul)
    313  = FIFFB_PROJ
        200  = FIFF_NCHAN (4b >i4)  = [306]
        3002 = FIFF_SPHERE_RADIUS (4b >i4)  = [1]
        3001 = FIFF_SPHERE_ORIGIN (12b >f4)  = [0. 0. 0.] ... array size=3
        314  = FIFFB_PROJ_ITEM
            206  = FIFF_COMMENT/FIFF_DESCRIPTION (24b str)  = ssp_68_magn.fif : PCA-v1 ... str len=24
            3411 = FIFF_PROJ_ITEM_KIND (4b >i4)  = [1]
            3417 = FIFF_PROJ_ITEM_CH_NAME_LIST (2447b str)  = MEG0111:MEG0112:MEG0113:MEG012 ... str len=2447
            3412 = FIFF_PROJ_ITEM_TIME (4b >f4)  = [0.]
            3414 = FIFF_PROJ_ITEM_NVEC (4b >i4)  = [1]
            3415 = FIFF_PROJ_ITEM_VECTORS (1236b >f4)  = [[-0.07794372  0.          0.  ... array size=306
        314  = FIFFB_PROJ_ITEM
            206  = FIFF_COMMENT/FIFF_DESCRIPTION (24b str)  = ssp_68_magn.fif : PCA-v2 ... str len=24
            3411 = FIFF_PROJ_ITEM_KIND (4b >i4)  = [1]
            3417 = FIFF_PROJ_ITEM_CH_NAME_LIST (2447b str)  = MEG0111:MEG0112:MEG0113:MEG012 ... str len=2447
            3412 = FIFF_PROJ_ITEM_TIME (4b >f4)  = [0.]
            3414 = FIFF_PROJ_ITEM_NVEC (4b >i4)  = [1]
            3415 = FIFF_PROJ_ITEM_VECTORS (1236b >f4)  = [[-0.07037429  0.          0.  ... array size=306
        (...)
        314  = FIFFB_PROJ_ITEM
            206  = FIFF_COMMENT/FIFF_DESCRIPTION (24b str)  = ssp_68_grad.fif : PCA-v2 ... str len=24
            3411 = FIFF_PROJ_ITEM_KIND (4b >i4)  = [1]
            3417 = FIFF_PROJ_ITEM_CH_NAME_LIST (2447b str)  = MEG0111:MEG0112:MEG0113:MEG012 ... str len=2447
            3412 = FIFF_PROJ_ITEM_TIME (4b >f4)  = [0.]
            3414 = FIFF_PROJ_ITEM_NVEC (4b >i4)  = [1]
            3415 = FIFF_PROJ_ITEM_VECTORS (1236b >f4)  = [[ 0.00000000e+00  4.50235838e ... array size=306

The tag 314 (FIFFB_PROJ_ITEM) contains the information of a single SSP. The tag 3411 (FIFF_PROJ_ITEM_KIND) indicates the type of projection. The tag 3417 (FIFF_PROJ_ITEM_CH_NAME_LIST) contains the list of channels to which the SSP is applied.

All MEGIN software work with the FIFF format.

Software

The available software from MEGIN are:

• DicomAccess is a DICOM v3.0 compliant application for transferring DICOM images. The images can be retrieved from a DICOM database or loaded from directories. The GET operation creates FIFF files from the single-frame 16-bit CT, enhanced CT, MR and enhanced MR images.

• Graph is a general purpose signal processor based on a LISP language interpreter. This tool is useful to visualize and compared datasets through its different layouts.

• GraphicsClipboard to generate screenshots of visualizations.

• MaxFilter to run (temporal) Signal-Space Separation[1][2]. See the section about MaxWell filter.

• MEG-MRI-Integration, also called mrilabl, is a tool to visualize volumetric data. It’s especially designed to combine MRI and MEG data.

• MRI-segmentation, also called seglab, is a segmentation program with both 2D processing operators (for single images) and 3D processing operators (for volume data).

• Plotting, also called xplotter, is a plotting tool mainly intended for plotting of evoked response data.

• SourceModelling, also called xfit, for dipole fitting.

• ViewBrain is used to visualize volumetric data. It can render segmented MRI as 3D images and overlay Equivalent Current Dipoles (ECDs).

Python

MNE-Python

FreeSurfer

MATLAB

The data analysis PC does not have a MATLAB license. However, applications using the MATLAB Runtime can be added upon request.

References

[1] (1,2)

Samu Taulu and Matti Kajola. Presentation of electromagnetic multichannel data: the signal space separation method. Journal of Applied Physics, 97(12):124905, 2005. doi:10.1063/1.1935742.

[2] (1,2)

Samu Taulu and Juha Simola. Spatiotemporal signal space separation method for rejecting nearby interference in MEG measurements. Physics in Medicine and Biology, 51(7):1759–1768, 2006. doi:10.1088/0031-9155/51/7/008.

[3] (1,2)

Mikko A. Uusitalo and Risto J. Ilmoniemi. Signal-space projection method for separating MEG or EEG into components. Medical & Biological Engineering & Computing, 35(2):135–140, 1997. doi:10.1007/BF02534144.