* Corresponding author.. E-mail address: hofmann@isip.mu-luebeck.de (U.G. Hofmann).  http://www.isip. mu-luebeck.de Neurocomputing 38}40 (2001) 1725}1734 Handling large "les of multisite microelectrode recordings for the European VSAMUEL consortium B. Weber, T. Malina, K. Menne, V. Metzler, A. Folkers, U.G. Hofmann * Institut fu ( r Signalverarbeitung und Proze}rechentechnik, Medizinische Universita ( t zu Lu ( beck, 23569 Lu ( beck, Germany Abstract The EU-funded project VSAMUEL aims to record from up to 128 channels from animal multi-electrode experiments. This will put a signi"cant load on data storage and retrieval. To decrease the "le size of 16 bit, 25 kHz recordings, compression algorithms were investigated and compared. Due to the high entropy of the signal, entropy coding was discarded and instead awaveletbasedcoderimplemented.Compressionratiosofupto1:20wereachieved,butonthe cost of loss of spike events. This was quanti"ed by a new spatio-temporal event detector, which takes the location of electrodes on our silicon probes into account.  2001 Elsevier Science B.V. All rights reserved. Keywords: Multi-site recording; Silicon probes; Wavelet; Compression; Morphological "lter; Image processing; Entropy 1. Introduction Modern neuroscience and our understanding of how the brain and its neural units work, is indisputably connected to the experimental ability of detecting their natural means of communication by electrical activity [6]. Even though extracellular record- ing from single cells is performed routinely, little is still known about network- and higher level activity. Intercepting this level of information processing requires the technological ability to place a high number of recording sites in close proximity to cells in question [18,20]. The use of standard microelectrodes or even tetrodes not 0925-2312/01/$-see front matter  2001 Elsevier Science B.V. All rights reserved. PII:S0925-2312(01)00538-0