Computers in Biology and Medicine 37 (2007) 1232 – 1240
www.intl.elsevierhealth.com/journals/cobm
A two-stage method for MUAP classification based on EMG decomposition
Christos D. Katsis
a,b
, Themis P. Exarchos
a,b
, Costas Papaloukas
c
, Yorgos Goletsis
d
,
Dimitrios I. Fotiadis
e,f ,∗
, Ioannis Sarmas
g
a
Department of Medical Physics, Medical School, University of Ioannina, GR 451 10 Ioannina, Greece
b
Unit of Medical Technology and Intelligent Information Systems, Department of Computer Science, University of Ioannina, P.O. Box 1186, GR 451 10
Ioannina, Greece
c
Department of Biological Applications and Technology, University of Ioannina, GR 451 10 Ioannina, Greece
d
Department of Economics, University of Ioannina, GR 451 10 Ioannina, Greece
e
Unit of Medical Technology and Intelligent Information Systems, Department of Computer Science, University of Ioannina, P.O. Box 1186, GR 451 10
Ioannina, Greece
f
Biomedical Research Institute—FORTH, GR 451 10 Ioannina, Greece
g
Department of Neurosurgery, Medical School, University of Ioannina, GR 451 10, Ioannina, Greece
Received 9 January 2006; accepted 6 November 2006
Abstract
A method for the extraction and classification of individual motor unit action potentials (MUAPs) from needle electromyographic signals is
presented. The proposed method automatically decomposes MUAPs and classifies them into normal, neuropathic or myopathic using a two-
stage feature-based classifier. The method consists of four steps: (i) preprocessing of EMG recordings, (ii) MUAP clustering and detection of
superimposed MUAPs, (iii) feature extraction and (iv) MUAP classification using a two-stage classifier. The proposed method employs Radial
Basis Function Artificial Neural Networks and decision trees. It requires minimal use of tuned parameters and is able to provide interpretation
for the classification decisions. The approach has been validated on real EMG recordings and an annotated collection of MUAPs. The success
rate for MUAP clustering is 96%, while the accuracy for MUAP classification is about 89%.
᭧ 2006 Elsevier Ltd. All rights reserved.
Keywords: Quantitative electromyography; Electromyogram decomposition; MUAP detection and classification; Radial basis function network; Decision trees
1. Introduction
Electromyography (EMG) is the study of the electrical ac-
tivity of the muscle and is a valuable tool in the assessment of
neuromuscular disorders. Computer-aided EMG has become an
indispensable tool in the daily activities of neurophysiology lab-
oratories in facilitating quantitative analysis and decision mak-
ing in clinical neurophysiology, rehabilitation, sport medicine
and human physiology. EMG findings are used to detect and
describe different disease processes affecting the Motor Unit
(MU), which is the smallest functional unit of the muscle. At
slight voluntary muscle contraction a motor unit action poten-
tial (MUAP) is recorded, reflecting the electrical activity of a
single anatomical MU [1]. MUAPs from different MUs tend to
∗
Corresponding author. Tel.: +30 2651098819; fax: +30 2651098889.
E-mail address: fotiadis@cs.uoi.gr (D.I. Fotiadis).
0010-4825/$ - see front matter
᭧
2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compbiomed.2006.11.010
have different shapes, which remain almost the same for each
discharge. Thus, MUAPs can be identified and tracked using
pattern recognition techniques. The resulting information can
be used to determine the origin of the disease, i.e. neuropathy
or myopathy [2–4].
When a patient maintains a low level of muscle contraction,
individual MUAPs can be easily recognized, since only a few
MUs are active. As contraction intensity increases, more MUs
are recruited; different MUAPs overlap, causing an interference
pattern (i.e. superimposed MUAPs) in which the neurophysi-
ologist cannot always detect individual MUAP shapes reliably.
The changes brought about by a particular disease alter the
functionality of the muscle and nerve cells, causing character-
istic changes in the MUAPs. Usually, in clinical EMG, neuro-
physiologists assess MUAPs from their shape using an oscillo-
scope and listening to their audio characteristics. Using these