Accurate discrimination of spike activity from extracellular recordings is very important when analysing neural activity. Analogue hardware discrimination is commonly used because of its speed and ease of use. However, such methods usually rely on ‘window discrimination’ that is dependent on the size of spikes compared with the background noise. Similar techniques are available in software packages that employ digital discrimination. However, both hardware and software are liable to miss spikes and count artifact events as biological spikes.

The methods proposed here combine the intuitive nature of oscilloscope window discrimination with digital shape evaluation. The first stage mimics a level discriminator: an initial trigger level set very close to the noise level is used to extract potential spikes and generous window discrimination can be used to exclude very large artifacts such as those of a stimulus. The second stage plots the real and imaginary coefficients of the fast Fourier transform of specifiable orders, for a user-selected region of potential spike shapes. Other parameters of shape, such as the maximum gradient or a cursor value at a given time reference, may also be plotted.

Event inclusion is specified by a graphical user interface based on set theory. A user-modifiable elliptical function can be used to include or exclude scatter points of three two-dimensional plots of customisable shape parameters. An intuitive colour-coded interface employs Boolean logic to enable the user to specify which subsets represent genuine spikes. To assist each decision, a waveform overlay is given and an inter-spike interval histogram is shown to highlight short intervals and missing spikes. The logical interface allows the user to review rapidly all those wave sections, possibly considered as artifacts, that were not counted as spikes. Prior to acceptance, the user is invited to review the whole waveform record with the discriminated extracts displayed alongside.

Missed spikes give inappropriately long intervals whereas artifacts counted as spikes can manifest as very short intervals. Accurate discrimination is essential when analysing inter-spike intervals rather than mean spike frequency. It is particularly important when attempting to distinguish different cells recorded simultaneously from the same electrode.

This work was supported by Merck, Sharp and Dohme, and the James Baird Fund.

DemonstrationsDemonstrationsDemonstrationsAccurate discrimination is essential when analysing inter-spike intervals rather than mean spike frequency. It is particularly important when attempting to distinguish different cells recorded simultaneously from the same electrode.

This work was supported by Merck, Sharp and Dohme, and the James Baird Fund.