A realistic nerve model was used to illustrate the clinical relevance of our findings. An FEM model of a cadaveric human femoral nerve, based on a histological nerve cross section [24 ], was chosen as an example of a realistic nerve model. A similar model is currently being used to improve the design of nerve cuff electrodes intended for femoral nerve stimulation [17 (link)]. A pair of neighboring fascicles was chosen to demonstrate the thp, Df, and neighboring fascicle effects in a realistic nerve model (Fig. 2 ). These two fascicles were chosen because of their differences in diameters and functions, which makes selective activation of these fascicles more desired for functional electrical stimulation. Fascicles 1 and 2 had diameters of 200 and 650 μm, respectively. Their centers were 550 μm apart.
Four FEM models were generated: 1), 2) a model of each individual fascicle in the pair, 3) a model of both fascicles in the pair, and 4) a model with all fascicles in the nerve cross section. Unlike our previous models, these models also included epineurium around the fascicles. Four perineurial thicknesses were tested, 0% and 3% of the each fascicle diameter, 30 μm (equal to 15% of fascicle 1 diameter and 4.6% of fascicle 2 diameter), and 50 μm (equal to 25% of fascicle 1 diameter and 7.7% of fascicle 2 diameter). These perineurial thicknesses were chosen based on thickness used in previous models [9 ], [12 (link)], [30 (link)] and our analysis of physiologic perineurial thickness (see Section III). The electrode contact was placed on the inner surface of the cuff and centered between the two fascicles in the chosen pair.
The percent activations of each of the two fascicles alone, with a neighboring fascicle, and with all fascicles in the cross section were compared over a range of recruitment levels (pulse widths of 0.01–0.1 ms). The effects of the Df and thp were also assessed based on the computed percent activations.
Four FEM models were generated: 1), 2) a model of each individual fascicle in the pair, 3) a model of both fascicles in the pair, and 4) a model with all fascicles in the nerve cross section. Unlike our previous models, these models also included epineurium around the fascicles. Four perineurial thicknesses were tested, 0% and 3% of the each fascicle diameter, 30 μm (equal to 15% of fascicle 1 diameter and 4.6% of fascicle 2 diameter), and 50 μm (equal to 25% of fascicle 1 diameter and 7.7% of fascicle 2 diameter). These perineurial thicknesses were chosen based on thickness used in previous models [9 ], [12 (link)], [30 (link)] and our analysis of physiologic perineurial thickness (see Section III). The electrode contact was placed on the inner surface of the cuff and centered between the two fascicles in the chosen pair.
The percent activations of each of the two fascicles alone, with a neighboring fascicle, and with all fascicles in the cross section were compared over a range of recruitment levels (pulse widths of 0.01–0.1 ms). The effects of the Df and thp were also assessed based on the computed percent activations.
