A new approach to spike sorting for multi-neuronal activities recorded with a tetrode—how ICA can be practical

doi:10.1016/S0168-0102(03)00103-2

Neuroscience Research

Volume 46, Issue 3, July 2003, Pages 265-272

https://doi.org/10.1016/S0168-0102(03)00103-2 Get rights and content

Abstract

Multi-neuronal recording with a tetrode is a powerful technique to reveal neuronal interactions in local circuits. However, it is difficult to detect precise spike timings among closely neighboring neurons because the spike waveforms of individual neurons overlap on the electrode when more than two neurons fire simultaneously. In addition, the spike waveforms of single neurons, especially in the presence of complex spikes, are often non-stationary. These problems limit the ability of ordinary spike sorting to sort multi-neuronal activities recorded using tetrodes into their single-neuron components. Though sorting with independent component analysis (ICA) can solve these problems, it has one serious limitation that the number of separated neurons must be less than the number of electrodes. Using a combination of ICA and the efficiency of ordinary spike sorting technique (k-means clustering), we developed an automatic procedure to solve the spike-overlapping and the non-stationarity problems with no limitation on the number of separated neurons. The results for the procedure applied to real multi-neuronal data demonstrated that some outliers which may be assigned to distinct clusters if ordinary spike-sorting methods were used can be identified as overlapping spikes, and that there are functional connections between a putative pyramidal neuron and its putative dendrite. These findings suggest that the combination of ICA and k-means clustering can provide insights into the precise nature of functional circuits among neurons, i.e. cell assemblies.

Introduction

Multi-neuronal recording has been used as a powerful electrophysiological technique to record the activities of multiple neurons simultaneously. Especially in recording from working brains, i.e. behaving animals, extracellular multi-neuronal recording can be used and signals from closely neighboring multiple neurons that construct a part of the local circuits can be detected with one electrode that has multiple recording tips near the neurons. The tetrode (Recce and O'keefe, 1989, Wilson and McNaughton, 1993) is an electrode constructed from four microwires insulated to the tips; it is one of the most reliable electrode for such extracellular multi-neuronal recording.

In order to identify the individual neuronal activities from the recorded multi-neuronal activity, techniques referred to as “spike sorting” are necessary (for a recent review see Lewicki, 1998). The ordinary spike sorting techniques use the different spike amplitudes and waveforms to classify single neuronal activities and such techniques are much needed when multi-neuronal activities are recorded using tetrodes. However, it is known that action potentials of individual neurons sometimes have varying shapes (Fee et al., 1996a). Moreover, we have no way of knowing whether some closely neighboring neurons generate simultaneous action potentials or not, because spike waveforms overlap on a common electrode when two or more neurons fire simultaneously. The overlapped waveforms cannot be separated by the ordinary spike sorting methods. These problems limit the ability to sort multi-neuronal activities recorded with tetrodes and the ability to detect precise spike timings (1–2 ms) among closely neighboring neurons around the microwires of the tetrode.

In this paper, we discuss the problems of the ordinary spike sorting methods and report a new procedure for spike sorting, by combining independent component analysis (ICA) (Comon, 1994) with an ordinary spike sorting technique; the new procedure is suitable for detecting the precise interactions among neighboring neurons.

Section snippets

Problems of ordinary spike sorting

There are many situations in which different neurons generate action potentials having very similar shapes in the recorded waveform. This happens when the neurons are similar in morphology and about equally distant from the different electrode tips. Moreover, it is well known that some extracellular recordings show the common occurrence of characteristic bursts called complex spikes. A stereotyped feature of these bursts is the progressive variation in spike shape during the burst. These

Automatic sorting using ICA and k-means clustering

To overcome both the non-stationarity and the spike-overlapping problems, we have developed an automatic procedure using a combination of ICA and k-means clustering.

The tetrode integrates both axonal and dendritic spikes

Due to the three-dimensional structure of the neuron, action potentials are not generated from a point source of the neuron, and extracellular electrodes may record the active backpropagation of dendritic action potentials. Several researchers, especially Buzsáki and colleagues, have suggested that the tetrode might be used to record action potentials not only from axons but also from dendrites by using a combination of intradendritic, intracellular and extracellular recordings in vivo (Buzsáki

Conclusions

The present procedure using a combination of ICA and k-means clustering can precisely detect single neurons even when they show brief bursts of high-frequency firing, which may reliably transmit signals to postsynaptic targets (Lisman, 1997) and have an increased possibility of generating overlapping spikes and non-stationary spike waveforms. The theory of synfire chains (Abeles, 1991), spike timing dependent plasticity (Bi and Poo, 1998) and the occurrence of dendritic backpropagating spikes,

Acknowledgements

This work was supported by the Sasakawa Scientific Research Grant to Dr Takahashi, and by the Special Coordination Funds for Promoting Science and Technology, the Research for the Future program, and RR21 program to Dr Sakurai.

References (36)

M. Abeles
Quantification, smoothing, and confidence limits for single-units’ histograms
J. Neurosci. Methods
(1982)
G.D. Brown et al.
Independent component analysis at the neural cocktail party
Trends Neurosci.
(2001)
P. Comon
Independent component analysis—a new concept
Sig. Proc.
(1994)
M.S. Fee et al.
Automatic sorting of multiple unit neuronal signals in the presence of anisotropic and non-Gaussian variability
J. Neurosci. Methods
(1996)
N.L. Golding et al.
Dendritic sodium spikes are variable triggers of axonal action potentials in hippocampal CA1 pyramidal neurons
Neuron
(1998)
C.M. Gray et al.
Tetrodes markedly improve the reliability and yield of multiple single-unit isolation from multi-unit recordings in cat striate cortex
J. Neurosci. Methods
(1995)
J.E. Lisman
Bursts as a unit of neural information: making unreliable synapses reliable
Trends Neurosci.
(1997)
Y. Sakurai
Population coding by cell assemblies—what it really is in the brain
Neurosci. Res.
(1996)
Y. Sakurai
Coding of temporal information by hippocampal individual cells and cell assemblies in the rat
Neuroscience
(2002)
G. Stuart et al.
Action potential initiation and backpropagation in neurons of the mammalian CNS
Trends Neurosci.
(1997)

S. Takahashi et al.

Classification of neuronal activities from tetrode recordings using independent component analysis

Neurocomputing

(2002)

M. Abeles

Corticonics: Neural Circuits of the Cerebral Cortex

(1991)

M. Abeles et al.

Multispike train analysis

Proc. IEEE

(1977)

M. Abeles et al.

Spatiotemporal firing patterns in the frontal cortex of behaving monkeys

J. Neurophysiol.

(1993)

G. Bi et al.

Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type

J. Neurosci.

(1998)

G. Buzsáki et al.

Pattern and inhibition-dependent invasion of pyramidal cell dendrites by fast spikes in the hippocampus in vitro

Proc. Natl. Acad. Sci. USA

(1996)

I. Cohen et al.

Contributions of intrinsic and synaptic activities to the generation of neuronal discharges in in vivo hippocampus

J. Physiol. (London)

(2000)

M.S. Fee et al.

Variability of extracellular spike waveforms of cortical neurons

J. Neurophysiol.

(1996)

Cited by (79)

Edge computing on TPU for brain implant signal analysis
2023, Neural Networks
The ever-increasing number of recording sites of silicon-based probes imposes a great challenge for detecting and evaluating single-unit activities in an accurate and efficient manner. Currently separate solutions are available for high precision offline evaluation and separate solutions for embedded systems where computational resources are more limited.
We propose a deep learning-based spike sorting system, that utilizes both unsupervised and supervised paradigms to learn a general feature embedding space and detect neural activity in raw data as well as predict the feature vectors for sorting. The unsupervised component uses contrastive learning to extract features from individual waveforms, while the supervised component is based on the MobileNetV2 architecture. One of the key advantages of our system is that it can be trained on multiple, diverse datasets simultaneously, resulting in greater generalizability than previous deep learning-based models.
We demonstrate that the proposed model does not only reaches the accuracy of current state-of-art offline spike sorting methods but has the unique potential to run on edge Tensor Processing Units (TPUs), specialized chips designed for artificial intelligence and edge computing. We compare our model performance with state of art solutions on paired datasets as well as on hybrid recordings as well. The herein demonstrated system paves the way to the integration of deep learning-based spike sorting algorithms into wearable electronic devices, which will be a crucial element of high-end brain–computer interfaces.
Enhanced Theta and High-Gamma Coupling during Late Stage of Rule Switching Task in Rat Hippocampus
2019, Neuroscience
Citation Excerpt :
We used one recording channel for LFP recording and the other three channels for unit recording. Recorded spike trains were sorted to isolate individual neuronal activities using independent component analysis (ICA) and k-means clustering called ICsort (Takahashi et al., 2003a, 2003b). We selected recording positions (pyramidal cell layer of the dorsal CA1) depending on the LFP power of the theta band and multiple single unit activities of putative pyramidal cells (Fig. 3B).
Hippocampal oscillations, particularly theta (6–12 Hz) and gamma (30–90 Hz) frequency bands, play an important role in several cognitive functions. Theta and gamma oscillations show cross-frequency coupling (CFC), wherein the phase of theta rhythm modulates the amplitude of the gamma oscillation, and this CFC is believed to reflect cell assembly dynamics in cognitive processes. Previous studies have reported that CFC strength correlates with the learning process. However, details on these dynamic correlations have not been elucidated. In the present study, we analyzed local field potentials recorded from the rat hippocampus during the learning of a rule-switching task. The modulation index, an index of the CFC strength, became higher in rule-guided behavior than in the no rule condition. The enhanced coupling between theta and high-gamma oscillations (60–90 Hz) changed during the late stage of learning. In contrast, the coupling between theta and low-gamma oscillations (30–60 Hz) did not show any changes during learning. These results suggest that the coupling between theta and gamma bands occurs during rule learning and that high- and low-gamma bands play different roles in rule switching.
Unsupervised neural spike sorting for high-density microelectrode arrays with convolutive independent component analysis
2016, Journal of Neuroscience Methods
Unsupervised identification of action potentials in multi-channel extracellular recordings, in particular from high-density microelectrode arrays with thousands of sensors, is an unresolved problem. While independent component analysis (ICA) achieves rapid unsupervised sorting, it ignores the convolutive structure of extracellular data, thus limiting the unmixing to a subset of neurons.
Here we present a spike sorting algorithm based on convolutive ICA (cICA) to retrieve a larger number of accurately sorted neurons than with instantaneous ICA while accounting for signal overlaps. Spike sorting was applied to datasets with varying signal-to-noise ratios (SNR: 3–12) and 27% spike overlaps, sampled at either 11.5 or 23 kHz on 4365 electrodes.
We demonstrate how the instantaneity assumption in ICA-based algorithms has to be relaxed in order to improve the spike sorting performance for high-density microelectrode array recordings. Reformulating the convolutive mixture as an instantaneous mixture by modeling several delayed samples jointly is necessary to increase signal-to-noise ratio. Our results emphasize that different cICA algorithms are not equivalent.
Spike sorting performance was assessed with ground-truth data generated from experimentally derived templates. The presented spike sorter was able to extract ≈90% of the true spike trains with an error rate below 2%. It was superior to two alternative (c)ICA methods (≈80% accurately sorted neurons) and comparable to a supervised sorting.
Our new algorithm represents a fast solution to overcome the current bottleneck in spike sorting of large datasets generated by simultaneous recording with thousands of electrodes.
Optical detection of neuron connectivity by random access two-photon microscopy
2016, Journal of Neuroscience Methods
Knowledge about the distribution, strength, and direction of synaptic connections within neuronal networks are crucial for understanding brain function. Electrophysiology using multiple electrodes provides a very high temporal resolution, but does not yield sufficient spatial information for resolving neuronal connection topology. Optical recording techniques using single-cell resolution have provided promise for providing spatial information. Although calcium imaging from hundreds of neurons has provided a novel view of the neural connections within the network, the kinetics of calcium responses are not fast enough to resolve each action potential event with high fidelity. Therefore, it is not possible to detect the direction of neuronal connections.
We took advantage of the fast kinetics and large dynamic range of the DiO/DPA combination of voltage sensitive dye and the fast scan speed of a custom-made random-access two-photon microscope to resolve each action potential event from multiple neurons in culture.
Long-duration recording up to 100 min from cultured hippocampal neurons yielded sufficient numbers of spike events for analyzing synaptic connections. Cross-correlation analysis of neuron pairs clearly distinguished synaptically connected neuron pairs with the connection direction.
The long duration recording of action potentials with voltage-sensitive dye utilized in the present study is much longer than in previous studies. Simultaneous optical voltage and calcium measurements revealed that voltage-sensitive dye is able to detect firing events more reliably than calcium indicators.
This novel method reveals a new view of the functional structure of neuronal networks.
Why sharing matters for electrophysiological data analysis
2015, Brain Research Bulletin
We present the case for the sharing of electrophysiological datasets and tools for their analysis. Some of the problems, both sociological and technical, associated with improving the sharing of data and analysis tools are discussed. The work that has been done to try to improve data and code sharing in the electrophysiology area is reviewed. The sharing aspects of the current large projects in brain research are considered.
A novel framework for feature extraction in multi-sensor action potential sorting
2015, Journal of Neuroscience Methods
Citation Excerpt :
However, temporal features are suboptimal for AP sorting since neurons of the same class located at roughly equal distances to the electrode can generate similar AP waveforms (Buzsáki, 2004), and therefore similar features. To overcome this problem, multi-sensor extracellular probes capable of recording time-aligned data from multiple spatial locations have been used (Harris et al., 2000; Takahashi et al., 2003; Emondi et al., 2004; Chelaru and Jog, 2005). The simplest way to extract features from multi-sensor measurements is to apply standard feature extraction techniques to individual channels, and then combine the extracted features into a single feature set.
Extracellular recordings of multi-unit neural activity have become indispensable in neuroscience research. The analysis of the recordings begins with the detection of the action potentials (APs), followed by a classification step where each AP is associated with a given neural source. A feature extraction step is required prior to classification in order to reduce the dimensionality of the data and the impact of noise, allowing source clustering algorithms to work more efficiently.
In this paper, we propose a novel framework for multi-sensor AP feature extraction based on the so-called Matched Subspace Detector (MSD), which is shown to be a natural generalization of standard single-sensor algorithms.
Clustering using both simulated data and real AP recordings taken in the locust antennal lobe demonstrates that the proposed approach yields features that are discriminatory and lead to promising results.
Unlike existing methods, the proposed algorithm finds joint spatio-temporal feature vectors that match the dominant subspace observed in the two-dimensional data without needs for a forward propagation model and AP templates.
The proposed MSD approach provides more discriminatory features for unsupervised AP sorting applications.

View all citing articles on Scopus

View full text

Update articleA new approach to spike sorting for multi-neuronal activities recorded with a tetrode—how ICA can be practical

Abstract

Introduction

Section snippets

Problems of ordinary spike sorting

Automatic sorting using ICA and k-means clustering

The tetrode integrates both axonal and dendritic spikes

Conclusions

Acknowledgements

J. Neurosci. Methods

Trends Neurosci.

Sig. Proc.

J. Neurosci. Methods

Neuron

J. Neurosci. Methods

Trends Neurosci.

Neurosci. Res.

Neuroscience

Trends Neurosci.

Neurocomputing

Corticonics: Neural Circuits of the Cerebral Cortex

Multispike train analysis

Proc. IEEE

Spatiotemporal firing patterns in the frontal cortex of behaving monkeys

J. Neurophysiol.

Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type

J. Neurosci.

Pattern and inhibition-dependent invasion of pyramidal cell dendrites by fast spikes in the hippocampus in vitro

Proc. Natl. Acad. Sci. USA

Contributions of intrinsic and synaptic activities to the generation of neuronal discharges in in vivo hippocampus

J. Physiol. (London)

Variability of extracellular spike waveforms of cortical neurons

J. Neurophysiol.

Update article
A new approach to spike sorting for multi-neuronal activities recorded with a tetrode—how ICA can be practical