| |
California State University, Electrical Engineering Department,
Long Beach, CA 90840, U.S.A.
The University of Iowa, Iowa Spine Research Center, Iowa
City, IA 52242, U.S.A.
Introduction: It has been observed that during sustained muscle
contractions, changes occur in the surface EMG signal. These
changes, known as the mean and the median frequency shifts (FS),
have been correlated to muscle fatigue. Zero-crossing (ZC) counting
has been also used for the same purpose. In this study, the aim
was to demonstrate a new processing technique for muscle fatigue
identification. This technique was based on discrete wavelet
transform (WT).
Materials and Methods: Five subject were randomly selected
for fatigue tests. Each subject performed one minute Sorenson
test three times, with 5 minute rest between each 3 1/3 minute
lifting tests. These tests were: (Test 1) standing without wearing
lumbar support, (Test 2) lifting a load without lumbar support,
and (Test 3) lifting a load with lumbar support. The lumbar support
used was modified belt designed to increase hoop stress along
the low back. The EMG signals were recorded by surface electrodes
placed on the spinae erector muscles (L3). the initial (0-4 sec)
and the final (56-60 sec EMG responses were recorded.
Results: and discussion: The normalized results of each tests
are as follows: The Mean/Median FS were, 1/1, 0.96/0.95, and
-0.14/-0.14, respectively. The ZC (initial - final) counted 1,
0.2, and -0.38. The signal energy differences calculated by WT
were 1, 0.04, and 0.5, indicating more fatigue in Test 2 and
less fatigue in Test 3.
Conclusion: The minimum FS was observed in Test 3 which matches
our findings with WT. The energy value obtained shows that less
fatigue was observed in Test 3.
Poster Session: European Spine Society 8th Annual Meeting,
September, 1997, Kos, Greece
|
