Add raw.rescale

doc/changes/devel/13018.newfeature.rst

@@ -0,0 +1 @@
+Add new :meth:`Raw.rescale <mne.io.Raw.rescale>` method to rescale the data in place, by `Clemens Brunner`_.

doc/sphinxext/related_software.py

@@ -81,6 +81,10 @@
         "Summary": "A graphical user interface for MNE",
     },
     # TODO: these do not set a valid homepage or documentation page on PyPI
+    "eeg_positions": {
+        "Home-page": "https://eeg-positions.readthedocs.io",
+        "Summary": "Compute and plot standard EEG electrode positions.",
+    },
     "mne-features": {
         "Home-page": "https://mne.tools/mne-features",
         "Summary": "MNE-Features software for extracting features from multivariate time series",  # noqa: E501

mne/conftest.py

@@ -178,6 +178,8 @@ def pytest_configure(config):
     ignore:__array__ implementation doesn't accept a copy.*:DeprecationWarning
     # quantities via neo
     ignore:The 'copy' argument in Quantity is deprecated.*:
+    # debugpy uses deprecated matplotlib API
+    ignore:The (non_)?interactive_bk attribute was deprecated.*:
     """  # noqa: E501
     for warning_line in warning_lines.split("\n"):
         warning_line = warning_line.strip()

mne/io/base.py

@@ -1500,6 +1500,67 @@ def resample(
             )
             return self, events
 
+    @verbose
+    def rescale(self, scale, *, verbose=None):
+        """Rescale channels.
+
+        .. warning::
+            MNE-Python assumes data are stored in SI base units. This function should
+            typically only be used to fix an incorrect scaling factor in the data to get
+            it to be in SI base units, otherwise unintended problems (e.g., incorrect
+            source imaging results) and analysis errors can occur.
+
+        Parameters
+        ----------
+        scale : int | float | dict
+            The scaling factor by which to multiply the data. If a float, the same
+            scaling factor is applied to all channels (this works only if all channels
+            are of the same type). If a dict, the keys must be valid channel types and
+            the values the scaling factors to apply to the corresponding channels.
+        %(verbose)s
+
+        Returns
+        -------
+        raw : Raw
+            The raw object with rescaled data.
+
+        Examples
+        --------
+        A common use case for EEG data is to convert from µV to V, since many EE
+        systems store data in µV, but MNE-Python expects the data to be in V. Therefore,
+        the data needs to be rescaled by a factor of 1e-6. To rescale all channels from
+        µV to V, you can do::
+
+            >>> raw.rescale(1e-6)  # doctest: +SKIP
+
+        Note that the previous example only works if all channels are of the same type.
+        If there are multiple channel types, you can pass a dict with the individual
+        scaling factors. For example, to rescale only EEG channels, you can do::
+
+            >>> raw.rescale({"eeg": 1e-6})  # doctest: +SKIP
+        """
+        _validate_type(scale, (int, float, dict), "scale")
+        _check_preload(self, "raw.rescale")
+
+        if isinstance(scale, int | float):
+            if len(self.get_channel_types(unique=True)) == 1:
+                self.apply_function(lambda x: x * scale, channel_wise=False)
+            else:
+                raise ValueError(
+                    "If scale is a scalar, all channels must be of the same type. "
+                    "Consider passing a dict instead."
+                )
+        else:
+            for ch_type, ch_scale in scale.items():
+                if ch_type not in self.get_channel_types():
+                    raise ValueError(f"Invalid channel type: {ch_type}")
+                else:
+                    self.apply_function(
+                        lambda x: x * ch_scale, picks=ch_type, channel_wise=False
+                    )
+
+        return self
+
     @verbose
     def crop(self, tmin=0.0, tmax=None, include_tmax=True, *, verbose=None):
         """Crop raw data file.

mne/io/tests/test_raw.py

@@ -1063,3 +1063,20 @@ def test_last_samp():
     raw = read_raw_fif(raw_fname).crop(0, 0.1).load_data()
     last_data = raw._data[:, [-1]]
     assert_array_equal(raw[:, -1][0], last_data)
+
+
+def test_rescale():
+    """Test rescaling channels."""
+    raw = read_raw_fif(raw_fname, preload=True)  # multiple channel types
+
+    with pytest.raises(ValueError, match="If scale is a scalar, all channels"):
+        raw.rescale(2)  # need to use dict
+
+    orig = raw.get_data(picks="eeg")
+    raw.rescale({"eeg": 2})  # need to use dict
+    assert_allclose(raw.get_data(picks="eeg"), orig * 2)
+
+    raw.pick("mag")  # only a single channel type "mag"
+    orig = raw.get_data()
+    raw.rescale(4)  # a scalar works
+    assert_allclose(raw.get_data(), orig * 4)

tools/circleci_dependencies.sh

@@ -19,4 +19,4 @@ python -m pip install --upgrade --progress-bar off \
     mne-icalabel mne-lsl mne-microstates mne-nirs mne-rsa \
     neurodsp neurokit2 niseq nitime openneuro-py pactools \
     plotly pycrostates pyprep pyriemann python-picard sesameeg \
-    sleepecg tensorpac yasa meegkit
+    sleepecg tensorpac yasa meegkit eeg_positions