FIX: Correctly set the calibration factor in Nihon Kohden reader (#13468)

Co-authored-by: Eric Larson <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: 3 people

doc/changes/dev/13468.bugfix.rst

@@ -0,0 +1 @@
+Correctly set the calibration factor in Nihon Kohden reader (which affects channel amplitudes), by `Tom Ma`_.

mne/datasets/config.py

@@ -87,7 +87,7 @@
 # update the checksum in the MNE_DATASETS dict below, and change version
 # here: ↓↓↓↓↓↓↓↓
 RELEASES = dict(
-    testing="0.166",
+    testing="0.167",
     misc="0.27",
     phantom_kit="0.2",
     ucl_opm_auditory="0.2",
@@ -115,7 +115,7 @@
 # Testing and misc are at the top as they're updated most often
 MNE_DATASETS["testing"] = dict(
     archive_name=f"{TESTING_VERSIONED}.tar.gz",
-    hash="md5:273c5919cf74198a39146e9cbc146ce0",
+    hash="md5:d82318a83b436ca2c7ca8420487c05c2",
     url=(
         "https://codeload.github.com/mne-tools/mne-testing-data/"
         f"tar.gz/{RELEASES['testing']}"

mne/io/nihon/nihon.py

@@ -162,9 +162,14 @@ def _read_21e_file(fname):
                 break
         else:
             warn(
-                f"Could not decode 21E file as one of {_encodings}; "
+                f"Could not decode {e_fname} as one of {_encodings}; "
                 f"Default channel names are chosen."
             )
+    else:
+        warn(
+            f"Could not find {e_fname} containing channel definitions; "
+            f"Default channel names are chosen."
+        )
 
     return _chan_labels
 
@@ -280,7 +285,7 @@ def _read_nihon_header(fname):
                     "Cannot read NK file with different sfreq in each datablock"
                 )
 
-    return header
+    return header, _chan_labels
 
 
 def _read_event_log_block(fid, t_block, version):
@@ -380,13 +385,13 @@ def _map_ch_to_type(ch_name):
     return "eeg"
 
 
-def _map_ch_to_specs(ch_name):
+def _map_ch_to_specs(ch_name, chan_labels_upper):
     unit_mult = 1e-3
     phys_min = -12002.9
     phys_max = 12002.56
     dig_min = -32768
-    if ch_name.upper() in _default_chan_labels:
-        idx = _default_chan_labels.index(ch_name.upper())
+    if ch_name.upper() in chan_labels_upper:
+        idx = chan_labels_upper.index(ch_name.upper())
         if (idx < 42 or idx > 73) and idx not in [76, 77]:
             unit_mult = 1e-6
             phys_min = -3200
@@ -432,7 +437,9 @@ def __init__(self, fname, preload=False, *, encoding="utf-8", verbose=None):
         data_name = fname.name
         logger.info(f"Loading {data_name}")
 
-        header = _read_nihon_header(fname)
+        # chan_labels are electrode codes defined in the .21E file.
+        # It is not the same as header["ch_names"].
+        header, chan_labels = _read_nihon_header(fname)
         metadata = _read_nihon_metadata(fname)
 
         # n_chan = len(header['ch_names']) + 1
@@ -447,8 +454,9 @@ def __init__(self, fname, preload=False, *, encoding="utf-8", verbose=None):
         if "meas_date" in metadata:
             with info._unlock():
                 info["meas_date"] = metadata["meas_date"]
-        chs = {x: _map_ch_to_specs(x) for x in info["ch_names"]}
 
+        chan_labels_upper = [x.upper() for x in chan_labels]
+        chs = {x: _map_ch_to_specs(x, chan_labels_upper) for x in info["ch_names"]}
         cal = np.array([chs[x]["cal"] for x in info["ch_names"]], float)[:, np.newaxis]
         offsets = np.array([chs[x]["offset"] for x in info["ch_names"]], float)[
             :, np.newaxis

mne/io/nihon/tests/test_nihon.py

@@ -30,7 +30,7 @@ def test_nihon_eeg():
     raw_edf.drop_channels(["Events/Markers"])
 
     assert raw._data.shape == raw_edf._data.shape
-    assert raw.info["sfreq"] == raw.info["sfreq"]
+    assert raw.info["sfreq"] == raw_edf.info["sfreq"]
     # a couple of ch names differ in the EDF
     edf_ch_names = {"EEG Mark1": "$A2", "EEG Mark2": "$A1"}
     raw_edf.rename_channels(edf_ch_names)
@@ -48,7 +48,7 @@ def test_nihon_eeg():
         raw = read_raw_nihon(fname_edf, preload=True)
 
     with pytest.raises(ValueError, match="Not a valid Nihon Kohden EEG file"):
-        raw = _read_nihon_header(fname_edf)
+        header, _ = _read_nihon_header(fname_edf)
 
     bad_fname = data_path / "eximia" / "text_eximia.nxe"
 
@@ -90,3 +90,38 @@ def return_channel_duplicates(fname):
     )
     with pytest.warns(RuntimeWarning, match=msg):
         read_raw_nihon(fname)
+
+
+@testing.requires_testing_data
+def test_nihon_calibration():
+    """Test handling of calibration factor and range in Nihon Kohden EEG files."""
+    fname = data_path / "NihonKohden" / "DA00100E.EEG"
+    raw = read_raw_nihon(fname, preload=True, encoding="cp936")
+
+    Fp1_idx = raw.ch_names.index("Fp1")
+    M1_idx = raw.ch_names.index("M1")
+    M2_idx = raw.ch_names.index("M2")
+
+    Fp1_info = raw.info["chs"][Fp1_idx]
+    M1_info = raw.info["chs"][M1_idx]
+    M2_info = raw.info["chs"][M2_idx]
+
+    # M1, M2 are EEG channels, just like Fp1.
+    # So they should have the same calibration factor and physical range.
+    assert_allclose(M1_info["cal"], Fp1_info["cal"])
+    assert_allclose(M2_info["cal"], Fp1_info["cal"])
+    assert_allclose(M1_info["range"], Fp1_info["range"])
+    assert_allclose(M2_info["range"], Fp1_info["range"])
+
+    fname_edf = data_path / "NihonKohden" / "DA00100E.EDF"
+    raw_edf = read_raw_edf(fname_edf, preload=True)
+    raw_edf.drop_channels(["Events/Markers"])
+    # a couple of ch names differ in the EDF
+    edf_ch_names = {"EEG Mark1": "$M1", "EEG Mark2": "$M2"}
+    raw_edf.rename_channels(edf_ch_names)
+
+    assert raw.ch_names == raw_edf.ch_names
+    assert raw._data.shape == raw_edf._data.shape
+    assert raw.info["sfreq"] == raw_edf.info["sfreq"]
+
+    assert_allclose(raw.get_data(), raw_edf.get_data())