Additional fixes based on review

sappelhoff · Apr 28, 2021 · b62e4cf · b62e4cf
1 parent 6aedaea
commit b62e4cf
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/docs/matlab_differences.rst b/docs/matlab_differences.rst
@@ -22,7 +22,7 @@ Differences in Signal Detrending
 --------------------------------
 
 In the PREP pipeline, trends (i.e., slow drifts in EEG baseline signal) are
-temporarily removed from the data prior to line-noise removal with CleanLine
+temporarily removed from the data prior to adaptive line-noise removal
 as well as prior to bad channel detection via :class:`~pyprep.NoisyChannels`,
 which occurs at multiple points during robust re-referencing. This is done to
 improve the accuracy of both of these processes, which are sensitive to

diff --git a/pyprep/utils.py b/pyprep/utils.py
@@ -134,10 +134,11 @@ def _eeglab_create_highpass(cutoff, srate):
 
     """
     TRANSITION_WIDTH_RATIO = 0.25
-    trans_bandwidth = cutoff if cutoff < 2 else cutoff * TRANSITION_WIDTH_RATIO
+    HAMMING_CONSTANT = 3.3  # note: not entirely clear what this represents
 
     # Calculate parameters for constructing filter
-    order = 3.3 / (trans_bandwidth / srate)
+    trans_bandwidth = cutoff if cutoff < 2 else cutoff * TRANSITION_WIDTH_RATIO
+    order = HAMMING_CONSTANT / (trans_bandwidth / srate)
     order = int(np.ceil(order / 2) * 2)  # ensure order is even
     stop = cutoff - trans_bandwidth
     transition = (stop + cutoff) / srate