Microelectrode arrays which could record approximately in nerve cell size from large nerve cell clusters encourages visual prosthesis design and application studies. Choosing useful information among recorded neural waveforms' analysis precisely with least computational complexity is of critical importance for compactness, energy efficiency of the electronics placed in intraocular region and data reduction for communication bandwidth. As a first step of neural signal analysis, spike identification phase is performed offline in silico using neural waveforms which is obtained from in vitro rabbit retina ganglion cell layer. It is concluded that custom made approach including thresholding followed by labeling process used for ease to use and less complexity is sufficient for neural signals with high signal to noise ratio.