Strategies to enhance FES-induced leg cycle ergometry for individuals with SCI.

摘要

Functional electrical stimulation-induced leg cycle ergometry (FES-LCE) has been shown to provide therapeutic benefits for individuals with spinal cord injury (SCI). However, its "one-size-fits-all" design is faced with a number of shortcomings. Purpose. To study strategies to improve FES-LCE exercise for individuals with SCI. Specific aims were to (1) compare pedaling dynamics of persons with SCI with able-bodied individuals; (2) identify stimulation patterns that minimize individual muscle force requirements; (3) measure force contribution of thigh and shank muscles involved in pedaling at different stimulation intensities and crank positions; and (4) develop, test, and compare a Probably Approximately Correct (PAC) model as a predictor of thigh muscle strength sufficient for FES-LCE with other well-known statistical methods. Methodology. Six male SCI subjects between ages 20--54 years (mean 32+/-12 years) participated in FES-LCE. Twelve able-bodied subjects between ages 18--50 years (mean 27+/-12 years) participated in LCE without FES. Subjects performed variable resistance leg cycling at a relatively constant cadence of 50 revolutions per minute. Motion capture system recorded joint kinematics of hip, knee, ankle, pedal, and crank. Orthogonal pedal forces were recorded using a piezoelectric force sensor. Surface electromyography (EMG) was measured electrical activity of shank (gastrocnemius, tibialis anterior) and thigh (hamstrings, quadriceps, and gluteus maximus) muscles in able-bodied subjects. PAC model was developed to identify thigh muscle strengths associated with FES-LCE performance. Results. EMG activities were different from the FES-LCE stimulation scheme. Comparisons in joint kinematics between able-bodied and SCI subjects showed dissimilar ankle motions at flywheel resistances greater than zero. Stimulation of quadriceps in SCI subjects produced greatest net positive crank torque across fixed crank positions and stimulation intensities (27.8Nm) and gluteus maximus produced smallest (1.9Nm). PAC model classified 100% of training set and 90.8% of validation set. Conclusions. These results indicate that (1) major changes in stimulation patterns of FES-LCE are necessary to improve pedaling effectiveness; (2) PAC learning may be an appropriate choice for identifying whether a prospective rider's thigh muscle strength is sufficient for FES-LCE. It is believed that these findings may be beneficial for the next generation of ES-induced leg cycle ergometry systems.