Lumbar muscle activity prediction under dynamic sagittal plane lifting conditions: Physiological and biomechanical modeling considerations.

Abstract

Understanding the effects of loading on the lumbar spine caused by manual handling of objects remains of significant human and economic concern. As muscle forces are often the largest contributors to the spinal load, this dissertation focuses on the general question of whether torso muscle activity during dynamic sagittal plane lifting can be predicted. As part of the study, the response of select torso and femoral muscles to static loads applied at the lumbar spine and across the pelvis was performed to help identify the sources of antagonistic torso muscle recruitment. Neither lumbar shear forces nor moments about the pelvis were good predictors of torso muscle antagonism. The results provide evidence that omission of these effects in biomechanical lumbar modeling probably is not introducing significant error in torso muscle activity prediction. To validate the muscle tension predictions during dynamic motions using EMG, the length-tension and force-velocity relations for the torso extensors were investigated. A technique using a general iEMG to muscle force mapping function was used to estimate the relationships from dynamic loading experiments with five subjects. The results revealed a linearly increasing length-tension relation as the torso flexes (r$\sp2$ = 0.47) and an approximately linear decrease in tension capability for increased extension speed (r$\sp2$ = 0.25). Large individual subject differences were found. Subject postures were dynamically measured using a sonic-based 3D tracking system. This system used an error reducing formulation of the triangulation equations that improved the accuracy of the system by 30% on average as compared with the traditional trigonometric method. Finally, a cortical-physiology based muscle prediction construct using a probability-like distributed moment histogram (DMH) was developed. Conceptually novel, it proposes that muscles are recruited in response to a dynamically modulated Gaussian distribution of the load moment direction and magnitude. This method was compared against an optimization formulation under dynamic sagittal plane lifting conditions. Both methods predicted erector spinae activity in good agreement with observed EMG (r$\sp2$ = 0.91). Latissimus dorsi predictions differed in magnitude by approximately 50% (DMH lower) and were not as well correlated (r$\sp2$ $\approx$ 0.4). Antagonistic external oblique activity was predicted (r$\sp2$ $\approx$ 0.07, st.err. 1.6% MVC) only by the DMH method.