Fix ICLabel topographic features on ICA fitted with channel selection (…

…#68)

* fix channel selection for topoplot

* fix style

* fix tests

* add test

* use testing instead of sample dataset

* fix style

* add entry to changelog

* add gh id

doc/whats_new.rst

@@ -32,6 +32,7 @@ Bug
 - Fix shape of ``'y_pred_proba'`` output from :func:`mne_icalabel.label_components` by `Mathieu Scheltienne`_ (:gh:`36`)
 - Add a warning if the ICA decomposition provided does not match the expected decomposition by ``ICLabel``  by `Mathieu Scheltienne`_ (:gh:`42`)
 - Fix extraction of PSD feature from ``ICLabel`` model on epochs by `Mathieu Scheltienne`_ (:gh:`64`)
+- Fix ICLabel topographic features on ICA fitted with a channel selection performed by ``picks`` by `Mathieu Scheltienne`_ (:gh:`68`)
 
 API
 ~~~

mne_icalabel/iclabel/features.py

@@ -1,4 +1,4 @@
-from typing import Tuple, Union
+from typing import List, Tuple, Union
 
 import numpy as np
 from mne import BaseEpochs
@@ -76,7 +76,7 @@ def get_iclabel_features(inst: Union[BaseRaw, BaseEpochs], ica: ICA):
     icaact = _compute_ica_activations(inst, ica)
 
     # compute topographic feature (float32)
-    topo = _eeg_topoplot(inst, icawinv)
+    topo = _eeg_topoplot(inst, icawinv, ica.ch_names)
 
     # compute psd feature (float32)
     psd = _eeg_rpsd(inst, ica, icaact)
@@ -168,12 +168,13 @@ def _compute_ica_activations(inst: Union[BaseRaw, BaseEpochs], ica: ICA) -> NDAr
 
 
 # ----------------------------------------------------------------------------
-def _eeg_topoplot(inst: Union[BaseRaw, BaseEpochs], icawinv: NDArray[float]) -> NDArray[float]:
+def _eeg_topoplot(
+    inst: Union[BaseRaw, BaseEpochs], icawinv: NDArray[float], picks: List[str]
+) -> NDArray[float]:
     """Topoplot feature."""
-    # TODO: Selection of channels is missing.
     ncomp = icawinv.shape[-1]
     topo = np.zeros((32, 32, 1, ncomp))
-    rd, th = _mne_to_eeglab_locs(inst)
+    rd, th = _mne_to_eeglab_locs(inst, picks)
     th = np.pi / 180 * th  # convert degrees to radians
     for it in range(ncomp):
         temp_topo = _topoplotFast(icawinv[:, it], rd, th)

mne_icalabel/iclabel/tests/test_features.py

@@ -160,7 +160,7 @@ def test_topoplotFast(file, eeglab_result_file):
     # load ICA
     ica = read_ica_eeglab(file)
     # convert coordinates
-    rd, th = _mne_to_eeglab_locs(inst)
+    rd, th = _mne_to_eeglab_locs(inst, ica.ch_names)
     th = np.pi / 180 * th
     # get icawinv
     icawinv, _ = _retrieve_eeglab_icawinv(ica)
@@ -190,7 +190,7 @@ def test_eeg_topoplot(file, eeglab_result_file):
     # get icawinv
     icawinv, _ = _retrieve_eeglab_icawinv(ica)
     # compute feature
-    topo = _eeg_topoplot(inst, icawinv)
+    topo = _eeg_topoplot(inst, icawinv, ica.ch_names)
     # load from eeglab
     topo_eeglab = loadmat(eeglab_result_file)["topo"]
     # compare

mne_icalabel/iclabel/tests/test_label_components.py

@@ -1,15 +1,15 @@
 import numpy as np
 import pytest
-from mne import create_info, make_fixed_length_epochs
-from mne.datasets import sample
+from mne import create_info, make_fixed_length_epochs, pick_types
+from mne.datasets import testing
 from mne.io import RawArray, read_raw
 from mne.preprocessing import ICA
 
 from mne_icalabel.iclabel import iclabel_label_components
 
-directory = sample.data_path() / "MEG" / "sample"
-raw = read_raw(directory / "sample_audvis_raw.fif", preload=False)
-raw.crop(0, 100).pick_types(eeg=True, exclude="bads")
+directory = testing.data_path() / "MEG" / "sample"
+raw = read_raw(directory / "sample_audvis_trunc_raw.fif", preload=False)
+raw.pick_types(eeg=True, exclude=[])
 raw.load_data()
 # preprocess
 raw.filter(l_freq=1.0, h_freq=100.0)
@@ -18,21 +18,27 @@
 
 @pytest.mark.filterwarnings("ignore::RuntimeWarning")
 @pytest.mark.parametrize(
-    "inst",
+    "inst, exclude",
     (
-        raw,
-        raw.copy().crop(0, 10),
-        raw.copy().crop(0, 1),
-        make_fixed_length_epochs(raw, duration=0.5),
-        make_fixed_length_epochs(raw, duration=1),
-        make_fixed_length_epochs(raw, duration=5),
-        make_fixed_length_epochs(raw, duration=10),
+        (raw, "bads"),
+        (raw.copy().crop(0, 8), "bads"),
+        (raw.copy().crop(0, 1), "bads"),
+        (make_fixed_length_epochs(raw, duration=0.5, preload=True), "bads"),
+        (make_fixed_length_epochs(raw, duration=1, preload=True), "bads"),
+        (make_fixed_length_epochs(raw, duration=5, preload=True), "bads"),
+        (raw, []),
+        (raw.copy().crop(0, 8), []),
+        (raw.copy().crop(0, 1), []),
+        (make_fixed_length_epochs(raw, duration=0.5, preload=True), []),
+        (make_fixed_length_epochs(raw, duration=1, preload=True), []),
+        (make_fixed_length_epochs(raw, duration=5, preload=True), []),
     ),
 )
-def test_label_components(inst):
+def test_label_components(inst, exclude):
     """Check that label_components does not raise on various data shapes."""
+    picks = pick_types(raw.info, eeg=True, exclude=exclude)
     ica = ICA(n_components=5, method="picard", fit_params=dict(ortho=False, extended=True))
-    ica.fit(inst)
+    ica.fit(inst, picks=picks)
     labels = iclabel_label_components(inst, ica)
     assert labels.shape == (ica.n_components_, 7)
 

mne_icalabel/iclabel/tests/test_utils.py

@@ -40,7 +40,7 @@ def test_loc(file, eeglab_result_file):
     when loading the datasets."""
     type_ = str(Path(file).stem)[-3:]
     inst = reader[type_](file, **kwargs[type_])
-    rd, th = _mne_to_eeglab_locs(inst)
+    rd, th = _mne_to_eeglab_locs(inst, picks=inst.ch_names)
     eeglab_loc = loadmat(eeglab_result_file)["loc"][0, 0]
     eeglab_rd = eeglab_loc["rd"]
     eeglab_th = eeglab_loc["th"]

mne_icalabel/iclabel/utils.py

@@ -5,7 +5,7 @@
 from numpy.typing import ArrayLike, NDArray
 
 
-def _mne_to_eeglab_locs(raw: BaseRaw) -> Tuple[NDArray[float], NDArray[float]]:
+def _mne_to_eeglab_locs(raw: BaseRaw, picks: List[str]) -> Tuple[NDArray[float], NDArray[float]]:
     """Obtain EEGLab-like spherical coordinate from EEG channel positions.
 
     TODO: @JACOB:
@@ -18,6 +18,8 @@ def _mne_to_eeglab_locs(raw: BaseRaw) -> Tuple[NDArray[float], NDArray[float]]:
     raw : mne.io.BaseRaw
         Instance of raw object with a `mne.montage.DigMontage` set with
         ``n_channels`` channel positions.
+    picks : list of str
+        List of channel names to include.
 
     Returns
     -------
@@ -44,7 +46,7 @@ def _cart2sph(_x, _y, _z):
         return azimuth, elevation, r
 
     # get the channel position dictionary
-    montage = raw.get_montage()
+    montage = raw.copy().pick_channels(picks, ordered=True).get_montage()
     positions = montage.get_positions()
     ch_pos = positions["ch_pos"]
 

mne_icalabel/tests/test_label_components.py

@@ -1,13 +1,13 @@
 import pytest
-from mne.datasets import sample
+from mne.datasets import testing
 from mne.io import read_raw
 from mne.preprocessing import ICA
 
 from mne_icalabel import label_components
 
-directory = sample.data_path() / "MEG" / "sample"
-raw = read_raw(directory / "sample_audvis_raw.fif", preload=False)
-raw.crop(0, 10).pick_types(eeg=True, exclude="bads")
+directory = testing.data_path() / "MEG" / "sample"
+raw = read_raw(directory / "sample_audvis_trunc_raw.fif", preload=False)
+raw.pick_types(eeg=True, exclude="bads")
 raw.load_data()
 # preprocess
 raw.filter(l_freq=1.0, h_freq=100.0)