Tweet
When using the linear envelope representation of the EMG signal, cyclical
motions (eg. gait) usually exhibit more consistent behaviour than
noncyclical (eg. shoulder abduction). This can be explained with a
continuity argument; there is a strong tendency to position the legs
properly for the next cycle in gait (since if not, the next cycle will be
uncomfortably long or short), but the arm has no such constraint on it at
the start or finish of an abduction movement.
The absence of such a constraint in noncyclical motion leads to awkward
end-effects (frequently leading to effects through the entire cycle), and
obtaining a useful average ("ensemble") within-subject linear envelope
can be difficult.
Is anyone else involved in a study of (general or specific) noncyclical
motion using linear envelopes? I'm interested in what problems you've
encountered and how you've dealt with them. Suggestions for reading
material would be helpful as well. I'm a math/stats person with some
signal processing knowledge, and my current work is development of
statistical methodology for analysis of linear envelopes.