Assessing New Techniques for Spike Detection on MEA Data

Spike detection and sorting on electrophysiological signals from MEA electrodes is demanding because noise levels are high, and many neurons may be simultaneously recorded by each electrode. In such data, there are always detection failures (false negatives and false positives). We present a new spike detection technique based on Cepstrum of Bispectrum (CoB) and assess it on real neural data. The technique is compared with four established techniques using simultaneously recorded intracellular and extracellular signals. The new technique outperforms existing techniques on detecting spikes in the extracellular signal which are due to recorded intracellular spikes.