Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Robust, automatic spike sorting using mixtures of multivariate t-distributions.

Shy Shoham1, Matthew R Fellows, Richard A Normann

  • 1Department of Bioengineering, University of Utah, Salt Lake City, UT 84112, USA. sshoham@princeton.edu

Journal of Neuroscience Methods
|August 9, 2003
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Thermal bioeffects considerations for transcranial focused ultrasound associated neurodegeneration.

Brain stimulation·2026
Same author

Rapid temporal processing in the olfactory bulb underlies concentration-invariant odor identification and signal decorrelation.

Nature neuroscience·2026
Same author

Bessel beam side lobe suppression via non-degenerate two-photon excitation.

Biomedical optics express·2026
Same author

Temporal sequence geometry enables odor recognition and generalization.

bioRxiv : the preprint server for biology·2026
Same author

CaBLAM: a high-contrast bioluminescent Ca<sup>2+</sup> indicator derived from an engineered Oplophorus gracilirostris luciferase.

Nature methods·2025
Same author

Brain-wide hemodynamic responses to precise transcranial ultrasound neuromodulation.

Brain stimulation·2025

Neural waveform classification using Gaussian mixture models is inaccurate. A new method using multivariate t-distributions and expectation-maximization reliably classifies neural activity, even with noisy data.

Area of Science:

  • Computational Neuroscience
  • Signal Processing
  • Statistical Modeling

Background:

  • Current methods for classifying neural activity often use Gaussian models.
  • Gaussian models fail to accurately represent the complex statistics of neural waveform data.
  • This limitation hinders precise analysis of neural recordings.

Purpose of the Study:

  • To demonstrate the limitations of Gaussian models for neural waveform statistics.
  • To introduce and validate the use of multivariate t-distributions for improved modeling.
  • To develop a robust algorithm for neural activity classification using these distributions.

Main Methods:

  • Analysis of neural waveform data to identify statistical distributions.
  • Adaptation of an expectation-maximization algorithm for mixture decomposition.

Related Experiment Videos

  • Application of the algorithm to t-distribution mixture models for neural data.

    • Main Results:

    • Data confirm non-Gaussian statistics in neural waveform samples.
    • Multivariate t-distributions provide a superior fit compared to Gaussian models.
    • The developed algorithm accurately decomposes t-distributions and classifies neural units.

    Conclusions:

    • The multivariate t-distribution offers a more accurate statistical model for neural waveforms.
    • The novel expectation-maximization algorithm reliably classifies neural units from complex data.
    • This approach enhances the accuracy of automatic neural activity classification systems.