Estimation of Peripheral Nerve Conduction Velocity Distributions

A complete and accurate description of the conduction properties of the entire population of nerve fibers within a peripheral nerve bundle has long been an important objective of clinical neuromuscular electrophysiology. Access to such information would increase the sensitivity and specificity of neuromuscular electrodiagnostics, and provide improved means of assessing nerve development, growth, regeneration, and response to treatment. One approach has been to generate an estimate of the distribution of conduction velocities (CVD) of all fibers by solving the inverse problem in electroneurography. The complexity and the inconvenience of current CVD estimation techniques have restricted their use to the research laboratory. In order to be accepted and utilized in a clinical setting, conduction velocity distribution estimates should be noninvasive and fast, without the need for complex procedures or costly equipment. A new technique for improving the estimation of conduction velocity distributions over a short segment of nerve is described and its feasibility is investigated. Using a short segment of nerve can significantly simplify the placement of surface electrodes while reducing the time and the discomfort involved in the acquisition of evoked compound action potential signals. The technique involves normalizing nerve-to-electrode transfer functions at a plurality of closely spaced recording sites. Following normalization, differences between compound action potential waveforms can be attributed to the effects of temporal dispersion and more accurate CVD estimates can be calculated from the short nerve segment. The feasibility of the method is evaluated using simulated and experimental compound action potential data. A formal study is performed to assess the performance of the technique in a clinical setting. Thesis Supervisor: Roger G. Mark Title: Distinguished Professor in Health Sciences and Technology; Professor of Electrical Engineering