Estimation of the conduction velocity distribution of peripheral nerve trunks.

The distribution of the conduction velocity (DCV) of peripheral nerve fibers was estimated from two compound action potentials (CAPs) that were evoked at two sites by electrical stimulation and recorded at a different site. In this paper, a regularized least-squares method with a smoothness constraint and self-adaptation of the regularization parameter was adopted. Computer simulation showed it to be effective, especially for noisy CAP signals. In addition, the conduction lengths of nerve fibers were automatically determined by a Fibonacci search, since the conduction lengths of nerve fibers could not be measured accurately due to slack in fibers or changes in length associated with the contraction of adjacent muscles.