Limits...
Fast, scalable, Bayesian spike identification for multi-electrode arrays.

Prentice JS, Homann J, Simmons KD, Tkačik G, Balasubramanian V, Nelson PC - PLoS ONE (2011)

Bottom Line: Our method can distinguish large numbers of distinct neural units, even when spikes overlap, and accounts for intrinsic variability of spikes from each unit.Human interaction plays a key role in our method; but effort is minimized and streamlined via a graphical interface.We illustrate our method on data from guinea pig retinal ganglion cells and document its performance on simulated data consisting of spikes added to experimentally measured background noise.

View Article: PubMed Central - PubMed

Affiliation: Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, Pennsylvania, United States of America. jprentic@sas.upenn.edu

ABSTRACT
We present an algorithm to identify individual neural spikes observed on high-density multi-electrode arrays (MEAs). Our method can distinguish large numbers of distinct neural units, even when spikes overlap, and accounts for intrinsic variability of spikes from each unit. As MEAs grow larger, it is important to find spike-identification methods that are scalable, that is, the computational cost of spike fitting should scale well with the number of units observed. Our algorithm accomplishes this goal, and is fast, because it exploits the spatial locality of each unit and the basic biophysics of extracellular signal propagation. Human interaction plays a key role in our method; but effort is minimized and streamlined via a graphical interface. We illustrate our method on data from guinea pig retinal ganglion cells and document its performance on simulated data consisting of spikes added to experimentally measured background noise. We present several tests demonstrating that the algorithm is highly accurate: it exhibits low error rates on fits to synthetic data, low refractory violation rates, good receptive field coverage, and consistency across users.

Show MeSH

Related in: MedlinePlus

Tests of the method.(A) (Top) The cumulative percentage of templates having false negative probabilities less than the indicated values. Error rates were measured in fits to synthetic data as the fraction of times a fit was not reported for a template when it was actually present. (Bottom) As above, but showing false positive probabilities (fraction of times a fit was reported for a template when it was not actually present). Results reported separately for fits to events with different numbers of overlapping spikes (inset colors). (B) Correlation in spike trains across fits by three different users (A, B, and C). Each curve corresponds to one pair of users and gives the cumulative fraction of templates having lower correlation than indicated. See main text for further details. (C) Cumulative fraction of templates having fewer refractory violations than indicated. Refractory violations are rare (see text). (D) The centers of 19 OFF cell receptive fields recorded from a single piece of tissue. To map a neuron's receptive field center, we first find the peak (in space and time) of the spike-triggered average stimulus. Restricting to the peak time, we apply cubic spline interpolation in space and then draw contour lines at 75% of the peak value.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3140468&req=5

pone-0019884-g003: Tests of the method.(A) (Top) The cumulative percentage of templates having false negative probabilities less than the indicated values. Error rates were measured in fits to synthetic data as the fraction of times a fit was not reported for a template when it was actually present. (Bottom) As above, but showing false positive probabilities (fraction of times a fit was reported for a template when it was not actually present). Results reported separately for fits to events with different numbers of overlapping spikes (inset colors). (B) Correlation in spike trains across fits by three different users (A, B, and C). Each curve corresponds to one pair of users and gives the cumulative fraction of templates having lower correlation than indicated. See main text for further details. (C) Cumulative fraction of templates having fewer refractory violations than indicated. Refractory violations are rare (see text). (D) The centers of 19 OFF cell receptive fields recorded from a single piece of tissue. To map a neuron's receptive field center, we first find the peak (in space and time) of the spike-triggered average stimulus. Restricting to the peak time, we apply cubic spline interpolation in space and then draw contour lines at 75% of the peak value.

Mentions: The template fitting algorithm was then run over this synthetic dataset and analyzed for false positive and false negative rates (Fig. 3A). We counted a false negative for a template every time that template was present in an event but not fit correctly to within ; we counted a false positive every time a template was fit to the data without actually being present. The error rates increased with the number of template overlaps; thus, for the fifty templates with amplitudes that exceed the noise, we separately plotted error rate histograms for each degree of overlap. Error rates were robustly low – even within extremely complex events with 5 overlapping spikes (rare in the data), the majority of spike templates had an error rate of a few percent or less. To gain perspective on these values, we measured the number of templates fit to each event in our recorded data: 60% of events contained 1 spike, 94% 3 or fewer spikes, and 98% had 5 spikes or fewer. Most of the errors were made on lower amplitude templates for which amplitude variations can lead to confusion with noise.


Fast, scalable, Bayesian spike identification for multi-electrode arrays.

Prentice JS, Homann J, Simmons KD, Tkačik G, Balasubramanian V, Nelson PC - PLoS ONE (2011)

Tests of the method.(A) (Top) The cumulative percentage of templates having false negative probabilities less than the indicated values. Error rates were measured in fits to synthetic data as the fraction of times a fit was not reported for a template when it was actually present. (Bottom) As above, but showing false positive probabilities (fraction of times a fit was reported for a template when it was not actually present). Results reported separately for fits to events with different numbers of overlapping spikes (inset colors). (B) Correlation in spike trains across fits by three different users (A, B, and C). Each curve corresponds to one pair of users and gives the cumulative fraction of templates having lower correlation than indicated. See main text for further details. (C) Cumulative fraction of templates having fewer refractory violations than indicated. Refractory violations are rare (see text). (D) The centers of 19 OFF cell receptive fields recorded from a single piece of tissue. To map a neuron's receptive field center, we first find the peak (in space and time) of the spike-triggered average stimulus. Restricting to the peak time, we apply cubic spline interpolation in space and then draw contour lines at 75% of the peak value.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3140468&req=5

pone-0019884-g003: Tests of the method.(A) (Top) The cumulative percentage of templates having false negative probabilities less than the indicated values. Error rates were measured in fits to synthetic data as the fraction of times a fit was not reported for a template when it was actually present. (Bottom) As above, but showing false positive probabilities (fraction of times a fit was reported for a template when it was not actually present). Results reported separately for fits to events with different numbers of overlapping spikes (inset colors). (B) Correlation in spike trains across fits by three different users (A, B, and C). Each curve corresponds to one pair of users and gives the cumulative fraction of templates having lower correlation than indicated. See main text for further details. (C) Cumulative fraction of templates having fewer refractory violations than indicated. Refractory violations are rare (see text). (D) The centers of 19 OFF cell receptive fields recorded from a single piece of tissue. To map a neuron's receptive field center, we first find the peak (in space and time) of the spike-triggered average stimulus. Restricting to the peak time, we apply cubic spline interpolation in space and then draw contour lines at 75% of the peak value.
Mentions: The template fitting algorithm was then run over this synthetic dataset and analyzed for false positive and false negative rates (Fig. 3A). We counted a false negative for a template every time that template was present in an event but not fit correctly to within ; we counted a false positive every time a template was fit to the data without actually being present. The error rates increased with the number of template overlaps; thus, for the fifty templates with amplitudes that exceed the noise, we separately plotted error rate histograms for each degree of overlap. Error rates were robustly low – even within extremely complex events with 5 overlapping spikes (rare in the data), the majority of spike templates had an error rate of a few percent or less. To gain perspective on these values, we measured the number of templates fit to each event in our recorded data: 60% of events contained 1 spike, 94% 3 or fewer spikes, and 98% had 5 spikes or fewer. Most of the errors were made on lower amplitude templates for which amplitude variations can lead to confusion with noise.

Bottom Line: Our method can distinguish large numbers of distinct neural units, even when spikes overlap, and accounts for intrinsic variability of spikes from each unit.Human interaction plays a key role in our method; but effort is minimized and streamlined via a graphical interface.We illustrate our method on data from guinea pig retinal ganglion cells and document its performance on simulated data consisting of spikes added to experimentally measured background noise.

View Article: PubMed Central - PubMed

Affiliation: Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, Pennsylvania, United States of America. jprentic@sas.upenn.edu

ABSTRACT
We present an algorithm to identify individual neural spikes observed on high-density multi-electrode arrays (MEAs). Our method can distinguish large numbers of distinct neural units, even when spikes overlap, and accounts for intrinsic variability of spikes from each unit. As MEAs grow larger, it is important to find spike-identification methods that are scalable, that is, the computational cost of spike fitting should scale well with the number of units observed. Our algorithm accomplishes this goal, and is fast, because it exploits the spatial locality of each unit and the basic biophysics of extracellular signal propagation. Human interaction plays a key role in our method; but effort is minimized and streamlined via a graphical interface. We illustrate our method on data from guinea pig retinal ganglion cells and document its performance on simulated data consisting of spikes added to experimentally measured background noise. We present several tests demonstrating that the algorithm is highly accurate: it exhibits low error rates on fits to synthetic data, low refractory violation rates, good receptive field coverage, and consistency across users.

Show MeSH
Related in: MedlinePlus