Comparative evaluation of motor unit architecture models.

We present a statistical evaluation and comparison of the simulation outcomes of nine different motor unit architecture modeling approaches, which derive from combinations of four motor unit territory placement algorithms with two innervation pattern algorithms (one of the combinations allows for a double approach). We test how well the outcomes of these models agree with well-established physiological principles. Our results show that algorithms based on independent and uniformly distributed territory placement always lead to an unwanted edge effect consisting in a decay of the number of overlapping motor unit territories toward the edge of the muscle cross section. This edge effect interferes with the properties of the simulated motor units. On the other hand, controlled placement of motor units, so as to minimize the spatial variance of muscle fiber density (MFD), can result in simulated motor units which better reflect the empirical evidence currently available about motor unit architecture.