ERPLAB Studio Panels: Interpolate Channels - ucdavis/erplab GitHub Wiki
The Interpolate Channels panel allows you to interpolate bad channels using a spherical spline algorithm (recommended for datasets containing approximately 25 or more channels) or an inverse distance algorithm (recommended for datasets containing less than approximately 25 channels).
You specify which channel(s) should be interpolated by listing the channel number(s) in the Interpolated chan box. If you want to use a subset of the other channels to compute the interpolated values, selected the Ignored chans option and list the to-be-ignored channels in the corresponding box. This is often used to exclude channels that have a different reference from the to-be-interpolated channels (e.g., to exclude bipolar EOG channels).
For epoched data, selecting the Interpolate marked epochs option allows to you limit interpolation to epochs that have been marked as containing artifacts. This can be useful when a channel is only occasionally “bad”. In other words, you can use the Artifact Detection routines to mark epochs in which the voltage in a given channel shows evidence of artifacts and then interpolate those epochs.
Typically, you will set a unique flag during artifact detection for the artifact in the channel you want to interpolate. For example, imagine that channel 19 is occasionally bad, producing large voltage deflections during approximately 20% of epochs. You could apply the moving window peak-to-peak artifact detection to this channel, specifying that Flag 4 should be set when this artifact is detected (in addition to Flag 1, which is set for all artifacts). When you perform interpolation, you would specify that you want to interpolate channel 19 for epochs in which Flag 4 is set. The voltage for channel 19 will then be interpolated for those epochs (but only if channel 19 is marked with an artifact). After the interpolation is complete, the artifact marks will be automatically cleared from those epochs so they are not excluded from averaging.