DICS-beamformer-for-Brainstorm

This is a DICS beamformer implementation for the Brainstorm (BS) software package. The dynamic imaging of coherent sources (DICS) beamformer technique enables the study of cortical sources of oscillatory activation in frequency-domain (Gross et al., 2001). DICS is a linearly constrained minimum variance beamformer in the frequency domain. It estimates the covariance matrix to calculate the spatial filter using the sensor-level cross-spectral density (CSD) matrix and applies the filter to the sensor-level CSD to reconstruct the source-level CSDs of pairwise voxel activations, providing coherence measures between the source pairs.

Note that the new pipeline is available at the BS GitHub repository: https://github.com/brainstorm-tools/brainstorm3/blob/master/toolbox/process/functions/process_ft_sourceanalysis_dics.m. The following details are based on the older version of the pipeline.

This implementation has mainly focused on localizing induced activations due to task-MEG responses, e.g., an overt definition naming language experiment.

Before running, follow these steps:

• Add the FieldTrip toolbox to the Matlab path, e.g., ft_path = 'xx/fieldtrip_20190419'; addpath(ft_path); ft_defaults;
• Estimate headmodel: overlapping spheres for surface-based and MRI volume for volumetric-based source mapping;

To run the DICS-BF in BS:

1. Open Brainstorm
2. Add (preprocessed epoched) trial responses to the processing window,
   

3. Select the DICS-BF source modeling from the process selection/Source/FieldTrip: ft_souceanalysis DICS-BF, vXX, and Run,

4. Choose DICS-beamformer as the source modeling approach, and MEG (MEG-MAG, or MEG GRAD) as the sensor type,

5. Pipeline estimates time-frequency responses (sensor-space, average across all sensors),

6. Select the time interval of post-vs-pre responses, e.g., [-0.3,0;0.7,1.2] 7. Select the frequency of interest, e.g., f=22Hz; A dpss smoothing window of 4Hz is applied (by default, see `vy_fft`, line 656) to estimate the cross-spectral density (CSD) matrix. 8. Results (surface map) are stored in the last trial response.

9. A sample result, an auditory definition naming task, DICS-BF compared against a dynamic Statistical Parametric Maps (dSPM), broadband 0.1-28Hz, is provided below.

For further inquiries, please contact [email protected].

Cite

1. Gross J, Kujala J, Hamalainen M, Timmermann L, Schnitzler A, Salmelin R. Dynamic imaging of coherent sources: Studying neural interactions in the human brain. Proc Natl Acad Sci U S A. 2001;98(2):694–9.
2. Youssofzadeh, V., Stout, J., Ustine, C., Gross, W.L., Lisa, L., Humphries, C.J., Binder, J.R., Raghavan, M., 2020. Mapping language from MEG beta power modulations during auditory and visual naming, NeuroImage. Elsevier Inc. https://doi.org/10.1016/j.neuroimage.2020.117090

Updates

• 07/08/21, A new tutorial wiki BS page was created: https://neuroimage.usc.edu/brainstorm/Tutorials/DICS
• 07/08/21, Pipeline is officially added to BS github repository: https://github.com/brainstorm-tools/brainstorm3/blob/master/toolbox/process/functions/process_ft_sourceanalysis_dics.m
• 07/01/21, a new update was made. Script was reformatted to match the standards in Brainstorm.
• 06/06/21, a new version of the pipeline process_ft_sourceanalysis_dics.m was added to the repository. For convenience and compatibility, variable inputs were integrated into a BS option GUI interface. The new version should produce the same results as the old version process_ft_sourceanalysis_DICS_BF.m.