1
2
3
4
5
6
7
8
9
10
11
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
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
|
# Authors: Kostiantyn Maksymenko <kostiantyn.maksymenko@gmail.com>
# Alexandre Gramfort <alexandre.gramfort@inria.fr>
#
# License: BSD (3-clause)
import numpy as np
from numpy.testing import assert_allclose
import pytest
from scipy.linalg import norm
from mne import SourceEstimate
from mne import read_source_spaces
from mne.datasets import testing
from mne.utils import requires_sklearn
from mne.simulation import metrics
from mne.simulation.metrics import (cosine_score,
region_localization_error,
precision_score, recall_score,
f1_score, roc_auc_score,
peak_position_error,
spatial_deviation_error)
data_path = testing.data_path(download=False)
src_fname = data_path / 'subjects' / 'sample' / 'bem' / 'sample-oct-6-src.fif'
@testing.requires_testing_data
def test_uniform_and_thresholding():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert = [src[0]['vertno'][0:1], []]
data = np.array([[0.8, -1.]])
stc_true = SourceEstimate(data, vert, 0, 0.002, subject='sample')
stc_bad = SourceEstimate(data, vert, 0, 0.002, subject='sample')
stc_bad.vertices = [stc_bad.vertices[0]]
with pytest.raises(ValueError, match='same number of vertices'):
metrics._uniform_stc(stc_true, stc_bad)
threshold = 0.9
stc1, stc2 = metrics._thresholding(stc_true, stc_true, threshold)
assert_allclose(stc1._data, np.array([[0, -1.]]))
assert_allclose(stc2._data, np.array([[0, -1.]]))
assert_allclose(threshold, metrics._check_threshold(threshold))
threshold = '90'
with pytest.raises(ValueError, match='Threshold if a str.*'):
metrics._check_threshold(threshold)
@testing.requires_testing_data
def test_cosine_score():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:1], []]
vert2 = [src[0]['vertno'][1:2], []]
data1 = np.ones((1, 2))
data2 = data1.copy()
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est1 = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
stc_est2 = SourceEstimate(data2, vert1, 0, 0.002, subject='sample')
E_per_sample1 = cosine_score(stc_true, stc_est1)
E_unique1 = cosine_score(stc_true, stc_est1, per_sample=False)
E_per_sample2 = cosine_score(stc_true, stc_est2)
E_unique2 = cosine_score(stc_true, stc_est2, per_sample=False)
assert_allclose(E_per_sample1, np.zeros(2))
assert_allclose(E_unique1, 0., atol=1e-08)
assert_allclose(E_per_sample2, np.ones(2))
assert_allclose(E_unique2, 1., atol=1e-08)
@testing.requires_testing_data
@requires_sklearn
def test_region_localization_error():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:1], []]
vert2 = [src[0]['vertno'][1:2], []]
dist = norm(src[0]['rr'][vert1[0]] - src[0]['rr'][vert2[0]])
data1 = np.ones((1, 2))
data2 = np.array([[0.8, 1]])
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est1 = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
E_per_sample1 = region_localization_error(stc_true, stc_est1, src)
E_per_sample2 = region_localization_error(stc_true, stc_est1, src,
threshold='70%')
E_unique = region_localization_error(stc_true, stc_est1, src,
per_sample=False)
assert_allclose(E_per_sample1, [np.inf, dist])
assert_allclose(E_per_sample2, [dist, dist])
assert_allclose(E_unique, dist)
@testing.requires_testing_data
@requires_sklearn
def test_precision_score():
"""Test simulation metrics."""
from sklearn.exceptions import UndefinedMetricWarning
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:2], []]
vert2 = [src[0]['vertno'][1:3], []]
vert3 = [src[0]['vertno'][0:1], []]
data1 = np.ones((2, 2))
data2 = np.ones((2, 2))
data3 = np.array([[0.8, 1]])
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est1 = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
stc_est2 = SourceEstimate(data3, vert3, 0, 0.002, subject='sample')
E_unique1 = precision_score(stc_true, stc_est1, per_sample=False)
E_unique2 = precision_score(stc_true, stc_est2, per_sample=False)
with pytest.warns(UndefinedMetricWarning, match='no predicted samples'):
E_per_sample1 = precision_score(stc_true, stc_est2)
E_per_sample2 = precision_score(stc_true, stc_est2,
threshold='70%')
with pytest.raises(ValueError, match='0 and 1'):
precision_score(stc_true, stc_est2, threshold=2)
# ### Tests to add
assert_allclose(E_unique1, 0.5)
assert_allclose(E_unique2, 1.)
assert_allclose(E_per_sample1, [0., 1.])
assert_allclose(E_per_sample2, [1., 1.])
@testing.requires_testing_data
@requires_sklearn
def test_recall_score():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:2], []]
vert2 = [src[0]['vertno'][1:3], []]
vert3 = [src[0]['vertno'][0:1], []]
data1 = np.ones((2, 2))
data2 = np.ones((2, 2))
data3 = np.array([[0.8, 1]])
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est1 = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
stc_est2 = SourceEstimate(data3, vert3, 0, 0.002, subject='sample')
E_unique1 = recall_score(stc_true, stc_est1, per_sample=False)
E_unique2 = recall_score(stc_true, stc_est2, per_sample=False)
E_per_sample1 = recall_score(stc_true, stc_est2)
E_per_sample2 = recall_score(stc_true, stc_est2, threshold='70%')
with pytest.raises(TypeError, match='numeric'):
precision_score(stc_true, stc_est2, threshold=None)
# ### Tests to add
assert_allclose(E_unique1, 0.5)
assert_allclose(E_unique2, 0.5)
assert_allclose(E_per_sample1, [0., 0.5])
assert_allclose(E_per_sample2, [0.5, 0.5])
@testing.requires_testing_data
@requires_sklearn
def test_f1_score():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:2], []]
vert2 = [src[0]['vertno'][1:3], []]
vert3 = [src[0]['vertno'][0:1], []]
data1 = np.ones((2, 2))
data2 = np.ones((2, 2))
data3 = np.array([[0.8, 1]])
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est1 = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
stc_est2 = SourceEstimate(data3, vert3, 0, 0.002, subject='sample')
E_unique1 = f1_score(stc_true, stc_est1, per_sample=False)
E_unique2 = f1_score(stc_true, stc_est2, per_sample=False)
E_per_sample1 = f1_score(stc_true, stc_est2)
E_per_sample2 = f1_score(stc_true, stc_est2, threshold='70%')
assert_allclose(E_unique1, 0.5)
assert_allclose(E_unique2, 1. / 1.5)
assert_allclose(E_per_sample1, [0., 1. / 1.5])
assert_allclose(E_per_sample2, [1. / 1.5, 1. / 1.5])
@testing.requires_testing_data
@requires_sklearn
def test_roc_auc_score():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:4], []]
vert2 = [src[0]['vertno'][0:4], []]
data1 = np.array([[0., 0., 1, 1]]).T
data2 = np.array([[0.1, -0.4, 0.35, 0.8]]).T
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
score = roc_auc_score(stc_true, stc_est, per_sample=False)
assert_allclose(score, 0.75)
@testing.requires_testing_data
def test_peak_position_error():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:1], []]
vert2 = [src[0]['vertno'][0:2], []]
data1 = np.array([[1]])
data2 = np.array([[1, 1.]]).T
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
r_mean = 0.5 * (src[0]['rr'][vert2[0][0]] + src[0]['rr'][vert2[0][1]])
r_true = src[0]['rr'][vert2[0][0]]
score = peak_position_error(stc_true, stc_est, src, per_sample=False)
assert_allclose(score, norm(r_true - r_mean))
with pytest.raises(ValueError, match='must contain only one dipole'):
peak_position_error(stc_est, stc_est, src)
data2 = np.array([[0, 0.]]).T
stc_est = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
score = peak_position_error(stc_true, stc_est, src, per_sample=False)
assert_allclose(score, np.inf)
@testing.requires_testing_data
def test_spatial_deviation():
"""Test simulation metrics."""
src = read_source_spaces(src_fname)
vert1 = [src[0]['vertno'][0:1], []]
vert2 = [src[0]['vertno'][0:2], []]
data1 = np.array([[1]])
data2 = np.array([[1, 1.]]).T
stc_true = SourceEstimate(data1, vert1, 0, 0.002, subject='sample')
stc_est = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
std = np.sqrt(0.5 * (0 + norm(src[0]['rr'][vert2[0][1]] -
src[0]['rr'][vert2[0][0]])**2))
score = spatial_deviation_error(stc_true, stc_est, src,
per_sample=False)
assert_allclose(score, std)
data2 = np.array([[0, 0.]]).T
stc_est = SourceEstimate(data2, vert2, 0, 0.002, subject='sample')
score = spatial_deviation_error(stc_true, stc_est, src,
per_sample=False)
assert_allclose(score, np.inf)
|
