Tg. Mcnaughton et Kw. Horch, ACTION-POTENTIAL CLASSIFICATION WITH DUAL-CHANNEL INTRAFASCICULAR ELECTRODES, IEEE transactions on biomedical engineering, 41(7), 1994, pp. 609-616
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.