ACTION-POTENTIAL CLASSIFICATION WITH DUAL-CHANNEL INTRAFASCICULAR ELECTRODES

Using recordings of peripheral nerve activity made with carbon fiber i ntrafascicular electrodes, we compared the performance of three differ ent recording techniques (single channel, differential, and dual chann el) and four different unit classification methods (linear discriminan t analysis, template matching, a novel time amplitude windowing techni que, and neural networks) in terms of errors in waveform classificatio n and artifact rejection. Dual channel recording provided uniformly su perior unit separability, neural networks gave the lowest classificati on error rates, and template matching had the best artifact rejection performance.