Static optimization of muscle forces during gait in comparison to EMG-to-force processing approach

Authors: Heintz, S, Gutierrez-Farewik, E.M.
Document Type: Article
Pubstate: Published
Journal: Gait & Posture
Volume: 26   279-288
Year: 2007


Individual muscle forces evaluated from experimental motion analysis may be useful in mathematical simulation, but require additional musculoskeletal and mathematical modelling. A numerical method of static optimization was used in this study to evaluate muscular forces during gait. The numerical algorithm used was built on the basis of traditional optimization techniques, i.e., constrained minimization technique using the Lagrange multiplier method to solve for constraints. Measuring exact muscle forces during gait analysis is not currently possible. The developed optimization method calculates optimal forces during gait, given a specific performance criterion, using kinematics and kinetics from gait analysis together with muscle architectural data. Experimental methods to validate mathematical methods to calculate forces are limited. Electromyography (EMG) is frequently used as a tool to determine muscle activation in experimental studies on human motion. A method of estimating force from the EMG signal, the EMG-to-force approach, was recently developed by Bogey and colleagues (2005) and is based on normalization of activation during a maximum voluntary contraction to documented maximal muscle strength. This method was adapted in this study as a tool with which to compare static optimization during a gait cycle. Muscle forces from static optimization and from EMG-to-force muscle forces show reasonably good correlation in the plantarflexor and dorsiflexor muscles, but less correlation in the knee flexor and extensor muscles. Additional comparison of the mathematical muscle forces from static optimization to documented averaged EMG data reveals good overall correlation to patterns of evaluated muscular activation. This indicates that on an individual level, muscular force patterns from mathematical models can arguably be more accurate than from those obtained from surface EMG during gait, though magnitude must still be validated.