Comparison of the three filter algorithms for detection of electrically-evoked short-latency responses in retinal ganglion cells

Abstract

When retinal ganglion cells are stimulated ex vivo by electrical current, short latency responses (within 30 ms) and long latency responses (over 30 ms) are evoked. The short-latency response is suspected to be originated from the direct electrical stimulation. For optimal stimulation protocol of retinal prosthesis, this short-latency response is getting important more and more. In this paper, we compared and evaluated performance of the three algorithms for the short-latency spike detection; suppression of artifacts by local polynomial approximation (SALPA), moving average filter (MAF), and forward-reverse filter (FR filter). The SALPA and the FR filter showed better detection performance comparing with the MAF algorithm.