Page 310 - Biomedical Engineering and Design Handbook Volume 1, Fundamentals

CHAPTER 12

ELECTROMYOGRAPHY

AS A TOOL TO ESTIMATE
MUSCLE FORCES

Qi Shao and Thomas S. Buchanan
University of Delaware, Newark, Delaware

12.1 INTRODUCTION: HOW TO ESTIMATE 12.5 AN EXAMPLE 300
MUSCLE FORCES 287 12.6 LIMITATIONS AND FUTURE
12.2 THE EMG SIGNAL 288 DEVELOPMENT OF EMG-DRIVEN
12.3 PROCESSING THE EMG SIGNAL 294 MODELS 303
12.4 EMG-DRIVEN MODELS TO ESTIMATE REFERENCES 304
MUSCLE FORCES 295

12.1 INTRODUCTION: HOW TO ESTIMATE MUSCLE FORCES?

Knowledge of internal muscle forces during movements is of great importance for understanding
human neuromuscular control strategies, developing better rehabilitation regimens, and improving
the design of prosthesis for patients with neurological disorders. However, the human neuromuscu-
loskeletal system is complicated and different muscles are finely coordinated to accomplish various
tasks, which makes their study difficult.
Unfortunately, in vivo muscle force measurement is invasive and only practical in very few cases.
Additionally, the musculoskeletal system is indeterminate, having more muscles than necessary for a
unique solution. For this reason, optimization techniques have been employed to predict muscle forces
using a variety of cost functions. Linear optimization techniques were first used for numerical conve-
nience (Seireg and Arvikar, 1973; Crowninshield, 1978). These linear cost functions were found to be
insufficient, so nonlinear cost functions have been developed, assuming one constant underlying neu-
romuscular control strategy during movement (Pedotti et al., 1978; Crowninshield and Brand, 1981;
Dul et al., 1984; Li et al., 1999). Pedotti et al. (1978) used a sum of individual muscle forces and nor-
malized muscle forces as their nonlinear cost function. Crowninshield and Brand (1981) utilized mus-
cle endurance as a nonlinear cost function to mathematically predict individual muscle forces. Muscle
endurance was described by a sum of muscle stresses to the third power. Dul et al. (1984) developed
a nonlinear optimization algorithm based on minimizing muscle fatigue, which took into account
maximal muscle force and composition of slow and fast twitch fibers. Li et al. (1999) found that the
number of degrees of freedom involved in their optimization played an important role in prediction of
the recruitment of antagonistic muscles rather than the selection of a particular cost function. They
concluded that a properly formulated inverse dynamics optimization should balance the knee joint in
three orthogonal planes. Other studies using forward dynamics simulation optimize muscle excitation

287