Abstract
The aim of the study is to determine which of the existing myelinated mammalian nerve fibre models better fits experimental data resulting from electrical stimulation of the human optic nerve and from propagation velocity measured on primates. The macroscopic electric potential is computed in a 3D, inhomogeneous and anisotropic nerve model. The Chiu-Sweeney (CS) and the Schwarz-Wesselink (SW) membrane descriptions are then considered. Variations in parameters that are not well established (encapsulation-tissue thickness, nerve-fascicle conductivity, geometric and electrochemical fibre cable parameters) are taken into account. Results demonstrate that the SW model predictions are in better agreement with the experimental data than those of the CS model, although thresholds are still too high. When channel densities are varied, the SW model turns out to be more robust than the CS model. For a suitable leakage channel density value (about 8% of the original one), the SW model predicts a conduction velocity of 11.4 ms-1 and an excitation threshold of 0.055 mA (for 0.1 ms pulse duration), which is in very good agreement with experimental values (11 ms-1 and 0.055 mA). Potassium current in the SW model is necessary for stability. Introduction of a potassium-like current can restore stability in the CS system.