[Mne_analysis] DICS beamformer {Disarmed}

Fri Feb 21 15:01:43 EST 2020

        External Email - Use Caution        

Dear Eric,

Thank you for your email. Yes, I have checked the coregistration and MEG-MRI coregistration seems fine given the overlay figure, and the dSPM for the evoked response of sound onsets (although it looks like the PT activity is spreading toward posterior STS; is this something common in MEG recon? Or does it mean the coreg is actually poor?).

Best regards,
-SG
________________________________
From: mne_analysis-bounces at nmr.mgh.harvard.edu <mne_analysis-bounces at nmr.mgh.harvard.edu> on behalf of Eric Larson <larson.eric.d at gmail.com>
Sent: Friday, February 14, 2020 9:46 AM
To: Discussion and support forum for the users of MNE Software <mne_analysis at nmr.mgh.harvard.edu>
Subject: Re: [Mne_analysis] DICS beamformer

        External Email - Use Caution

Have you checked some basics? For example, do you have at some points sensor-space patterns that look like bilateral auditory activation? In your source reconstruction do you get a reasonable auditory N100 localization to the stimulus onset? I would check these first (if possible for your paradigm) if you haven't. Knowing that they are correct rules out things like coregistration or epoching errors and helps narrow it down to being related to the method of source localization (DICS).

Eric

On Thu, Feb 13, 2020 at 3:17 PM Seung Goo Kim, Ph.D. <seunggoo.kim at duke.edu<mailto:seunggoo.kim at duke.edu>> wrote:

        External Email - Use Caution

Dear list,

I am having difficulties in applying the DICS beamformer on induced responses by an amplitude modulation in auditory signals. The script I used was:

import mne, os, pandas as pd, numpy as np, matplotlib.pyplot as plt
from mne.time_frequency import csd_morlet
from mne.beamformer import make_dics, apply_dics_csd
mne.set_config("SUBJECTS_DIR", "/mnt/data/MEG_pitch/data/fs+mne")
df = pd.read_excel(
    "/mnt/data/MEG_pitch/data/rawdata/MEGStudy/Subjects.xlsx",
    sheet_name="rawdata")
subjects = df.SubjectID[(df.goodmeg_for_tsss * df.goodmri_for_coreg) == True]

subj=subjects[1]
dname_meg = '/mnt/data/MEG_pitch/data/fs+mne/' + subj + '/meg/'
epochs = mne.read_epochs(dname_meg + 'all_tsss_hp0.5Hz_notch50sHz_cor_epo.fif')
fname_fwd = "/mnt/data/MEG_pitch/data/fs+mne/" + subj + "/bem/" + subj + "-4098-fwd.fif"
fwd = mne.read_forward_solution(fname_fwd)
freq = [40]
n_cycles = 4
cond = 'CT20RNR'
epochs_cond = epochs[cond]

# compute cross-spectral density over the whole epoch

csd = csd_morlet(epochs_cond, tmin=-0.5, tmax=2.3, frequencies=freq,
                 decim=20, n_cycles=n_cycles)
filters = make_dics(epochs.info<https://urldefense.proofpoint.com/v2/url?u=http-3A__epochs.info&d=DwMGaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=Oo6kXvV9N4AHFS-wWXXAKKDl4dUe6z92y4XtIzLkZJY&m=03nUvJsvDMigS5h-APMr-YTrVTN8dzxfUVgR1MXBUkU&s=3YrE589SYu_GBEuvYVeamwSDpiNWdMNLZSRsyH4Nrvs&e=>, fwd, csd.mean(), pick_ori='max-power')

csd_base = csd_morlet(epochs_cond, tmin=-0.5, tmax=0, frequencies=freq,
                      decim=20, n_cycles=n_cycles)
baseline_power = apply_dics_csd(csd_base.mean(), filters)
csd_R = csd_morlet(epochs_cond, tmin=1.7, tmax=2.2, frequencies=freq,
                   decim=20, n_cycles=n_cycles)
R_power = apply_dics_csd(csd_R.mean(), filters)
stc = R_power[0] / baseline_power[0]

stc.plot(subject=subj, surface="inflated", hemi="both", time_viewer=False,
        initial_time=0, transparent=True, backend='mayavi', views='lat', size=400)

In the stimulus, only a specific part (from 1.4 sec to 2.3 sec after stimulus onset) has a regular amplitude modulation, so I wanted to compare this part vs. baseline (from -0.5 sec to 0 sec).

So what I did was (1) computing a common filter for the entire epoch (from -0.5 sec to 2.3 sec after stimulus onset), (2) computing CSD for the baseline (from -0.5 sec to 0 sec), (3) computing CSD for the "regular" part (from 1.7 sec to 2.2 sec; same 500 ms as the baseline), and (4) found the ratio of them.

Since this was an auditory experiment, I expected this should be localized in the auditory cortices (bilaterally), but the result looks completely unexpected:
[cid:17044277e7dcb971f161]
[cid:17044277e7dcb971f162]

The code is simply a modification of the tutorial (https://mne.tools/dev/auto_examples/inverse/plot_dics_source_power.html<https://urldefense.proofpoint.com/v2/url?u=https-3A__mne.tools_dev_auto-5Fexamples_inverse_plot-5Fdics-5Fsource-5Fpower.html&d=DwMGaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=Oo6kXvV9N4AHFS-wWXXAKKDl4dUe6z92y4XtIzLkZJY&m=03nUvJsvDMigS5h-APMr-YTrVTN8dzxfUVgR1MXBUkU&s=UJEk1ej1vp7FarL5-twGTh9sv8H0P3YAAbTbokjGRHw&e=>), so I am very puzzled.
Compute source power using DICS beamfomer — MNE 0.20.dev0 documentation<https://urldefense.proofpoint.com/v2/url?u=https-3A__mne.tools_dev_auto-5Fexamples_inverse_plot-5Fdics-5Fsource-5Fpower.html&d=DwMGaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=Oo6kXvV9N4AHFS-wWXXAKKDl4dUe6z92y4XtIzLkZJY&m=03nUvJsvDMigS5h-APMr-YTrVTN8dzxfUVgR1MXBUkU&s=UJEk1ej1vp7FarL5-twGTh9sv8H0P3YAAbTbokjGRHw&e=>
Compute source power using DICS beamfomer¶. Compute a Dynamic Imaging of Coherent Sources (DICS) 1 filter from single-trial activity to estimate source power across a frequency band. This example demonstrates how to source localize the event-related synchronization (ERS) of beta band activity in this dataset: Somatosensory
mne.tools

Any suggestions on where to look/check would be greatly appreciated!

Best,
Seung-Goo Kim
_______________________________________________
Mne_analysis mailing list
Mne_analysis at nmr.mgh.harvard.edu<mailto:Mne_analysis at nmr.mgh.harvard.edu>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis<https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.edu_mailman_listinfo_mne-5Fanalysis&d=DwMGaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=Oo6kXvV9N4AHFS-wWXXAKKDl4dUe6z92y4XtIzLkZJY&m=03nUvJsvDMigS5h-APMr-YTrVTN8dzxfUVgR1MXBUkU&s=r8ifHkkMe6YvBsKZ3EaN3AORFhcQYILVcbJSw891kxU&e=>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200221/4488045f/attachment-0001.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 122489 bytes
Desc: image.png
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200221/4488045f/attachment-0005.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 93833 bytes
Desc: image.png
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200221/4488045f/attachment-0006.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: snapshot.png
Type: image/png
Size: 350721 bytes
Desc: snapshot.png
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200221/4488045f/attachment-0007.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: S01-maxfilt-run0-head-dev-crop.png
Type: image/png
Size: 187303 bytes
Desc: S01-maxfilt-run0-head-dev-crop.png
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200221/4488045f/attachment-0008.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Figure_2.png
Type: image/png
Size: 178048 bytes
Desc: Figure_2.png
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200221/4488045f/attachment-0009.png 