|
|
Precision Decomposition Technique
for EMG Signals
Overview |
Introduction & History | Equipment |
Performance
Introduction & History
EMG Decomposition is the process of separating and
identifying the individual action potentials that comprise the
Electromyographic (EMG) signal.
A group led by C.J. De Luca has been developing a such a
technique for three decades. The Precision Decomposition I consists
of a set of algorithms that uses: template matching, template shape
tracking, firing statistics and superposition resolution to decompose three
(3) channels of EMG signals that are detected with a quadrifilar needle or a
quadrifilar wire electrode. The old version was able to decompose
automatically, but rarely reached accuracies above 70%. The files required
operator assisted editing to achieve greater accuracies that, for some
files, reached 100%. Refer to the publication [4] for
proof.
We are now developing a more advanced Precision Decomposition II
that is much faster and more accurate.
Precision Decomposition II utilizes the Artificial Intelligence
technology known as IPUS. Developed in collaboration with Dr. Nawab of the
IDEA lab at Boston University and V. Lesser in the 1990's, IPUS supports
integrated processing and understanding of signals.
The evolution milestones are categorized in the following table.
| 1972 |
First description of the quadrifilar, three channel, differential
EMG electrode |
[1] |
| 1978 |
- First public presentation at the 8th Neuroscience Meeting
- template matching in three dimension
- statistical prediction of expected firings
|
[2] |
| 1982 |
- First peer-reviewed scientific publications
- template updating to accommodate small changes in
the action potential shapes due to needle movement
- resolution of superpositions
- addition of operator assisted editor
|
[3a, 3b] |
| 1984 |
First proof of accuracy |
[4] |
| 1988 |
Artificial Intelligence algorithms included |
[5] |
| 1999 |
Modified for wire quadrifilar electrode |
[6] |
| 2002 |
First description of enhanced artificial intelligence
algorithms |
[7] |
| [1] |
De Luca CJ and Forrest J. An electrode for recording single motor
unit activity during strong muscle contractions, IEEE BME
Transactions, 19: 367-372, 1972 |
| [2] |
LeFever, RS and De Luca, C J. Decomposition of action potential
trains. Proceedings of 8th Annual Meeting of the Society for
Neuroscience 229, 1978 |
| [3a] |
LeFever RS, De Luca CJ. A procedure for decomposing the myoelectric
signal into its constituent action potentials. Part I. Technique, theory
and implementation. IEEE BME Transactions, 29: 149–157, 1982 |
| [3b] |
LeFever RS, Xenakis AP, De Luca CJ. A procedure for decomposing the
myoelectric signal into its constituent action potentials. Part II.
Execution and test for accuracy. IEEE BME Transactions, 29:
158–164, 1982 |
| [4] |
Mambrito B, De Luca CJ. A technique for the detection, decomposition
and analysis of the EMG signal. EEG Clin. Neurophysiol. 58:
175–188, 1984 |
| [5] |
Broman H. Knowledge-Based Signal Processing in the Decomposition of
Myoelectric Signals. IEEE Trans Eng Med Biol: 24-28, 1988 |
| [6] |
De Luca CJ and Adam A. Decomposition and analysis of intramuscular
electromyographic signals. In: Modern Techniques in Neuroscience
Research, edited by Windhorst U and Johansson H. Heidelberg: Springer,
1999, p. 757-776 |
| [7] |
Hochstein L, Nawab SH, and Wotiz R. An AI-Based Software
Architecture for a Biomedical Application. SCI-2002: Proceedings of
The 6'th World Multiconference on Systemics, Cybernetics and
Informatics, Orlando, July, 2002, Volume XI, pp 60 - 64 |
|
