Adaptation of the Two-CAP method for conduction velocity distribution estimation in multi-channel recordings

Closed-loop neural interfaces capable of both stimulating and recording from peripheral nerves have the potential to enhance the long-term efficacy of neural implants. One challenge associated with closed loop interfaces is the accurate estimation of the distribution of active fibre conduction velocities (DCV) when recording the immediate effect of stimulation. DCV estimation has been performed in monopolar surface recordings using the Two-CAP method. This work extends the Two-CAP method and demonstrates its application to bipolar in-vivo recordings made with multiple-electrode arrays. A sensitivity analysis was conducted using simulated data with ground truth to ascertain the stability and limits of the algorithm before experimental data was examined. The sensitivity analysis highlighted that recording distance shows a considerable impact on the performance of this extended Two-CAP method, as well as the velocity interval chosen when creating the model. The in-vivo data was also compared against an equivalent simulated model, and a relatively low mean squared error was obtained when comparing the two distributions.