RESUMEN
A method for automatically analyzing cortical signals recorded with voltage-sensitive dyes and a photodiode array is described. First, a forward-backward non-linear filtering technique is used to eliminate the background noise and preserve the fast transients of the signals. Then the filtered signals are deconvoluted from their maximal values by using a gaussian function. The different components of the signals can be identified and characterized by their respective latencies, amplitudes, plateau durations, and slopes. These parameters can be used for subsequent statistical analysis. This automated method is much faster than a manual analysis because of the large number of responses that are optically recorded. Moreover, it can be easily applied to different experimental protocols and to other signals such as field potentials.