Strategies for High-Performance Resource-Efficient Compression of Neural Spike Recordings
Johansson AJ (2014) Strategies for High-Performance Resource-Efficient Compression of Neural Spike
Recordings. PLoS ONE 9(4): e93779. doi:10.1371/journal.pone.0093779
Strategies for High-Performance Resource-Efficient Compression of Neural Spike Recordings
Palmi Thor Thorbergsson 0
Martin Garwicz 0
Jens Schouenborg 0
Anders J. Johansson 0
Maurice J. Chacron, McGill University, Canada
0 1 Department of Experimental Medical Science, Lund University, Lund, Sweden, 2 Department of Electrical and Information Technology, Lund University, Lund, Sweden, 3 Neuronano Research Center, Lund University , Lund , Sweden
Brain-machine interfaces (BMIs) based on extracellular recordings with microelectrodes provide means of observing the activities of neurons that orchestrate fundamental brain function, and are therefore powerful tools for exploring the function of the brain. Due to physical restrictions and risks for post-surgical complications, wired BMIs are not suitable for long-term studies in freely behaving animals. Wireless BMIs ideally solve these problems, but they call for low-complexity techniques for data compression that ensure maximum utilization of the wireless link and energy resources, as well as minimum heat dissipation in the surrounding tissues. In this paper, we analyze the performances of various system architectures that involve spike detection, spike alignment and spike compression. Performance is analyzed in terms of spike reconstruction and spike sorting performance after wireless transmission of the compressed spike waveforms. Compression is performed with transform coding, using five different compression bases, one of which we pay special attention to. That basis is a fixed basis derived, by singular value decomposition, from a large assembly of experimentally obtained spike waveforms, and therefore represents a generic basis specially suitable for compressing spike waveforms. Our results show that a compression factor of 99.8%, compared to transmitting the raw acquired data, can be achieved using the fixed generic compression basis without compromising performance in spike reconstruction and spike sorting. Besides illustrating the relative performances of various system architectures and compression bases, our findings show that compression of spikes with a fixed generic compression basis derived from spike data provides better performance than compression with downsampling or the Haar basis, given that no optimization procedures are implemented for compression coefficients, and the performance is similar to that obtained when the optimal SVD based basis is used.
-
Funding: This work was supported by a Linnaeus Grant from the Swedish Research Council (no. 60012701), a grant from the Knut and Alice Wallenberg
Foundation (no. 2004.0119) and the Medical and Engineering Faculties at Lund University. The funders had no role in study design, data collection and analysis,
decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
Brain-machine interfaces (BMIs) are systems that provide a
signal pathway between the central nervous system (CNS) and the
outside world and thereby a means of observing the neuronal
activity that underlies behavior. One class of BMIs employ
intracranially implanted microelectrodes to record the changes in
extracellular potential induced by activities of neurons
surrounding them [1]. The extracellular recording is composed of spikes
representing action potentials in near-by neurons, noise consisting
of spikes from distant neurons, local field potentials (LFPs)
representing synaptic activity and thermal noise generated in the
recording electronics [2].
By isolating the spiking components of the individual neurons
that contribute to the extracellular recording, the firing patterns of
those neurons can be characterized and correlated with events or
learning processes in the motor or sensory domains [3] to reveal
the dynamics of neuronal circuits that govern behavior. The
fundamental steps in this procedure are (a) spike detection and
extraction, (b) spike alignment and (c) spike sorting. Spike
detection is commonly based on detecting the local increase in
signal energy or amplitude associated with the firing of a spike and
its aim is to pinpoint the temporal occurrence of spike waveforms.
Spike alignment is important for the subsequent spike sorting step
and it involves shifting the detected spike waveforms in time to
have them aligned with respect to a given waveform landmark, e.g.
the point of maximum amplitude. Spike sorting typically involves
first extracting spike features (i.e. waveform characteristics that
ideally are the same for spikes coming from the same neuron but
different for spikes coming from different neurons) and then
classifying the spikes based on the extracted features [4].
BMIs are often implemented with multiple recording channels
(multielectrode arrays), which results (...truncated)