Linear classifier on sensor data with plot patterns and filters

Here decoding, a.k.a MVPA or supervised machine learning, is applied to M/EEG data in sensor space. Fit a linear classifier with the LinearModel object providing topographical patterns which are more neurophysiologically interpretable [1] than the classifier filters (weight vectors). The patterns explain how the MEG and EEG data were generated from the discriminant neural sources which are extracted by the filters. Note patterns/filters in MEG data are more similar than EEG data because the noise is less spatially correlated in MEG than EEG.

# Authors: Alexandre Gramfort <alexandre.gramfort@inria.fr>
#          Romain Trachel <trachelr@gmail.com>
#          Jean-Remi King <jeanremi.king@gmail.com>
#
# License: BSD-3-Clause
# Copyright the MNE-Python contributors.

from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler

import mne
from mne import EvokedArray, io
from mne.datasets import sample

# import a linear classifier from mne.decoding
from mne.decoding import LinearModel, Vectorizer, get_coef

print(__doc__)

data_path = sample.data_path()
sample_path = data_path / "MEG" / "sample"

Set parameters

raw_fname = sample_path / "sample_audvis_filt-0-40_raw.fif"
event_fname = sample_path / "sample_audvis_filt-0-40_raw-eve.fif"
tmin, tmax = -0.1, 0.4
event_id = dict(aud_l=1, vis_l=3)

# Setup for reading the raw data
raw = io.read_raw_fif(raw_fname, preload=True)
raw.filter(0.5, 25, fir_design="firwin")
events = mne.read_events(event_fname)

# Read epochs
epochs = mne.Epochs(
    raw, events, event_id, tmin, tmax, proj=True, decim=2, baseline=None, preload=True
)
del raw

labels = epochs.events[:, -1]

# get MEG data
meg_epochs = epochs.copy().pick(picks="meg", exclude="bads")
meg_data = meg_epochs.get_data(copy=False).reshape(len(labels), -1)

Opening raw data file /home/circleci/mne_data/MNE-sample-data/MEG/sample/sample_audvis_filt-0-40_raw.fif...
    Read a total of 4 projection items:
        PCA-v1 (1 x 102)  idle
        PCA-v2 (1 x 102)  idle
        PCA-v3 (1 x 102)  idle
        Average EEG reference (1 x 60)  idle
    Range : 6450 ... 48149 =     42.956 ...   320.665 secs
Ready.
Reading 0 ... 41699  =      0.000 ...   277.709 secs...
Filtering raw data in 1 contiguous segment
Setting up band-pass filter from 0.5 - 25 Hz

FIR filter parameters
---------------------
Designing a one-pass, zero-phase, non-causal bandpass filter:
- Windowed time-domain design (firwin) method
- Hamming window with 0.0194 passband ripple and 53 dB stopband attenuation
- Lower passband edge: 0.50
- Lower transition bandwidth: 0.50 Hz (-6 dB cutoff frequency: 0.25 Hz)
- Upper passband edge: 25.00 Hz
- Upper transition bandwidth: 6.25 Hz (-6 dB cutoff frequency: 28.12 Hz)
- Filter length: 993 samples (6.613 s)

[Parallel(n_jobs=1)]: Done  17 tasks      | elapsed:    0.0s
[Parallel(n_jobs=1)]: Done  71 tasks      | elapsed:    0.1s
[Parallel(n_jobs=1)]: Done 161 tasks      | elapsed:    0.2s
[Parallel(n_jobs=1)]: Done 287 tasks      | elapsed:    0.5s
Not setting metadata
145 matching events found
No baseline correction applied
Created an SSP operator (subspace dimension = 4)
4 projection items activated
Using data from preloaded Raw for 145 events and 76 original time points (prior to decimation) ...
0 bad epochs dropped

Decoding in sensor space using a LogisticRegression classifier

clf = LogisticRegression(solver="liblinear")  # liblinear is faster than lbfgs
scaler = StandardScaler()

# create a linear model with LogisticRegression
model = LinearModel(clf)

# fit the classifier on MEG data
X = scaler.fit_transform(meg_data)
model.fit(X, labels)

# Extract and plot spatial filters and spatial patterns
for name, coef in (("patterns", model.patterns_), ("filters", model.filters_)):
    # We fitted the linear model onto Z-scored data. To make the filters
    # interpretable, we must reverse this normalization step
    coef = scaler.inverse_transform([coef])[0]

    # The data was vectorized to fit a single model across all time points and
    # all channels. We thus reshape it:
    coef = coef.reshape(len(meg_epochs.ch_names), -1)

    # Plot
    evoked = EvokedArray(coef, meg_epochs.info, tmin=epochs.tmin)
    fig = evoked.plot_topomap()
    fig.suptitle(f"MEG {name}")

• 
• 

Let’s do the same on EEG data using a scikit-learn pipeline

X = epochs.pick(picks="eeg", exclude="bads")
y = epochs.events[:, 2]

# Define a unique pipeline to sequentially:
clf = make_pipeline(
    Vectorizer(),  # 1) vectorize across time and channels
    StandardScaler(),  # 2) normalize features across trials
    LinearModel(  # 3) fits a logistic regression
        LogisticRegression(solver="liblinear")
    ),
)
clf.fit(X, y)

# Extract and plot patterns and filters
for name in ("patterns_", "filters_"):
    # The `inverse_transform` parameter will call this method on any estimator
    # contained in the pipeline, in reverse order.
    coef = get_coef(clf, name, inverse_transform=True)
    evoked = EvokedArray(coef, epochs.info, tmin=epochs.tmin)
    fig = evoked.plot_topomap()
    fig.suptitle(f"EEG {name[:-1]}")

• 
• 

References

[1]

Stefan Haufe, Frank Meinecke, Kai Görgen, Sven Dähne, John-Dylan Haynes, Benjamin Blankertz, and Felix Bießmann. On the interpretation of weight vectors of linear models in multivariate neuroimaging. NeuroImage, 87:96–110, 2014. doi:10.1016/j.neuroimage.2013.10.067.