Identification of Spikes by Computer

Publisher Summary This chapter provides an overview of the factors that an investigator has to consider when selecting a commercial spike identification system or when designing a customized one. It describes the features of the program Spike Finder. Spike validation is a key operation that is often neglected. The times of occurrence of the spikes are usually to be used in some analysis subsequent to the spike identification itself. It is, therefore, indispensable to have some measurement of the errors involved in spike identification, so as not to reach erroneous conclusions in subsequent analysis of the time structure of the data. The more detection and classification are automated, or use complex features, the more this precaution is important to minimize errors. Spike analysis can be divided into two tasks: (1) detection and (2) classification. The task of spike detection is essentially that of separating spikes from noise, that is, from unwanted signals of any type. It is often based on the use of some sort of voltage threshold, a software equivalent of the window discriminators that for many years were the primary means of selecting specific spikes. If a threshold is the basis of spike detection, it follows that the larger the signal-to-noise ratio, the more reliable the ensuing spike detection will be. Even the best computer method will fail if the signals of interest are barely detectable.