Skip to content

Backus–Gilbert

Solver ID: BG

Usage

from invert import Solver

# fwd = ...    (mne.Forward object)
# evoked = ... (mne.Evoked object)

solver = Solver("BG")
solver.make_inverse_operator(fwd)
stc = solver.apply_inverse_operator(evoked)
stc.plot()

Overview

Resolution-optimizing linear inverse method that trades off spatial resolution versus noise amplification using Backus–Gilbert theory.

References

  1. Backus, G., & Gilbert, F. (1968). The resolving power of gross earth data. Geophysical Journal of the Royal Astronomical Society, 16(2), 169–205.

API Reference

Bases: BaseSolver

Class for the Backus Gilbert inverse solution.

Source code in invert/solvers/minimum_norm/backus_gilbert.py 
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
class SolverBackusGilbert(BaseSolver):
    """Class for the Backus Gilbert inverse solution."""

    meta = SolverMeta(
        acronym="BG",
        full_name="Backus–Gilbert",
        category="Minimum Norm",
        description=(
            "Resolution-optimizing linear inverse method that trades off spatial "
            "resolution versus noise amplification using Backus–Gilbert theory."
        ),
        references=[
            "Backus, G., & Gilbert, F. (1968). The resolving power of gross earth data. Geophysical Journal of the Royal Astronomical Society, 16(2), 169–205.",
        ],
    )

    def __init__(self, name="Backus-Gilbert", **kwargs):
        self.name = name
        return super().__init__(**kwargs)

    def make_inverse_operator(self, forward, *args, alpha="auto", **kwargs):
        """Calculate inverse operator.

        Parameters
        ----------
        forward : mne.Forward
            The mne-python Forward model instance.
        alpha : float
            The regularization parameter.

        Return
        ------
        self : object returns itself for convenience
        """
        super().make_inverse_operator(forward, *args, alpha=alpha, **kwargs)
        _, n_dipoles = self.leadfield.shape
        pos = pos_from_forward(forward, verbose=self.verbose)
        dist = cdist(pos, pos)

        W_BG = []
        for i in range(n_dipoles):
            W_gamma_BG = np.diag(dist[i, :])
            W_BG.append(W_gamma_BG)

        C = []
        for i in range(n_dipoles):
            C_gamma = self.leadfield @ W_BG[i] @ self.leadfield.T
            C.append(C_gamma)

        F = self.leadfield @ self.leadfield.T

        E = []
        for i in range(n_dipoles):
            E_gamma = C[i] + F
            E.append(E_gamma)

        L = self.leadfield @ np.ones((n_dipoles, 1))

        T = []
        for i in range(n_dipoles):
            E_gamma_pinv = np.linalg.pinv(E[i])
            T_gamma = (E_gamma_pinv @ L) / (L.T @ E_gamma_pinv @ L)
            T.append(T_gamma)

        inverse_operators = [
            np.stack(T, axis=0)[:, :, 0],
        ]

        self.inverse_operators = [
            InverseOperator(inverse_operator, self.name)
            for inverse_operator in inverse_operators
        ]
        return self

__init__ 

__init__(name='Backus-Gilbert', **kwargs)
Source code in invert/solvers/minimum_norm/backus_gilbert.py 
28
29
30
def __init__(self, name="Backus-Gilbert", **kwargs):
    self.name = name
    return super().__init__(**kwargs)

make_inverse_operator 

make_inverse_operator(
    forward, *args, alpha="auto", **kwargs
)

Calculate inverse operator.

Parameters:

Name Type Description Default
forward Forward

The mne-python Forward model instance.

 required
alpha float

The regularization parameter.

 'auto'
Return

self : object returns itself for convenience

Source code in invert/solvers/minimum_norm/backus_gilbert.py 
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
def make_inverse_operator(self, forward, *args, alpha="auto", **kwargs):
    """Calculate inverse operator.

    Parameters
    ----------
    forward : mne.Forward
        The mne-python Forward model instance.
    alpha : float
        The regularization parameter.

    Return
    ------
    self : object returns itself for convenience
    """
    super().make_inverse_operator(forward, *args, alpha=alpha, **kwargs)
    _, n_dipoles = self.leadfield.shape
    pos = pos_from_forward(forward, verbose=self.verbose)
    dist = cdist(pos, pos)

    W_BG = []
    for i in range(n_dipoles):
        W_gamma_BG = np.diag(dist[i, :])
        W_BG.append(W_gamma_BG)

    C = []
    for i in range(n_dipoles):
        C_gamma = self.leadfield @ W_BG[i] @ self.leadfield.T
        C.append(C_gamma)

    F = self.leadfield @ self.leadfield.T

    E = []
    for i in range(n_dipoles):
        E_gamma = C[i] + F
        E.append(E_gamma)

    L = self.leadfield @ np.ones((n_dipoles, 1))

    T = []
    for i in range(n_dipoles):
        E_gamma_pinv = np.linalg.pinv(E[i])
        T_gamma = (E_gamma_pinv @ L) / (L.T @ E_gamma_pinv @ L)
        T.append(T_gamma)

    inverse_operators = [
        np.stack(T, axis=0)[:, :, 0],
    ]

    self.inverse_operators = [
        InverseOperator(inverse_operator, self.name)
        for inverse_operator in inverse_operators
    ]
    return self