Distributed moment histogram: a neurophysiology based method of agonist and antagonist trunk muscle activity prediction.

Abstract

A neurocortical-based technique of muscle recruitment is presented to solve the muscle indeterminacy problem for lumbar torso modeling. Cortical recordings from behaving primates have established motor cortex cells that respond to a wide range of task directions, but are tuned to a preferred direction. A characteristic activity pattern of these neurons seems to be associated with effort direction. It was hypothesized that a model which recruits muscles based on a similar distribution would predict antagonistic muscle activity with greater realism than a widely referenced optimization formulation. The predictions of the Distributed Moment Histogram (DMH) method were evaluated under common speed (< 30 degrees s-1) sagittal plane lifting conditions using five subjects. The predicted forces showed high correspondence with agonist and antagonist myoelectric patterns. The mean coefficient of determination for the erector spinae was r2 = 0.91, and 0.41 for the latissimus. For the antagonistic muscles, the rectus abdominus was found to be electrically silent (< 3% MVC) and no activity was predicted by the method. The external oblique muscle was observed to be minimally active (< 16% MVC), and the DMH method predicted its mostly constant activity with a mean standard error of 1.6% MVC. The realistic antagonistic predictions supported the hypothesis and justify this cortical based technique as an alternative for muscle tension estimation in biomechanical torso modeling. A primary advantage of this method is its computational simplicity and direct physiologic analog.