1 / 15

Spike Sorting based on Dominant-Sets clustering

Spike Sorting based on Dominant-Sets clustering. Dimitrios A. Adamos PhD Candidate School of Biology, Aristotle University. What is spike sorting?. Spikes. Noise. Spike Sorting Algorithms. Buzsáki, G. (2004) . Large-scale recording of neuronal ensembles . Nature Neuroscience.

Télécharger la présentation

Spike Sorting based on Dominant-Sets clustering

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Spike Sorting based on Dominant-Sets clustering Dimitrios A. Adamos PhD Candidate School of Biology, Aristotle University Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  2. What is spike sorting? Spikes Noise Spike Sorting Algorithms Buzsáki, G. (2004). Large-scale recording of neuronal ensembles. Nature Neuroscience Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  3. Spike sorting in a nutshell Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  4. Spike sorting applications BrainGate Linderman et al. (2008). Signal processing challenges for neural prostheses. IEEE Signal Processing Magazine Hochberg et al. (2006). Neuronal ensemble control of prosthetic devices by a human with tetraplegia. Nature Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  5. Spike sorting applications • Functional STN Targeting during DBS surgery Wong et al. (2009). Functional localization and visualization of the STN from microelectrode recordings acquired during DBS surgery with unsupervised machine learning. J Neural Eng Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  6. Open problems in spike sorting #1 • Resolution of overlapping spikes Adamos DA, Laskaris NA , Kosmidis EK and Theophilidis G. NASS: An empirical approach to Spike Sorting with overlap resolution based on a hybrid Noise-Assisted methodology (2010) Journal of Neuroscience Methods Article in Press doi:10.1016/j.jneumeth.2010.04.018 Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  7. Open problems in spike sorting #2 • Goal of this study: Correct estimation of active neurons Challenges: Noise & Sparsely firing neurons Common clustering errors: Under-clustering & over-clustering Under-clustering example Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  8. Methods • Combination of two methods from the graph-theoretic domain #1 ISOMAP Graph-theoretic feature extraction #2 Dominant-sets Graph-theoretic clustering Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  9. Methods #1: Non-linear low-dimensional representation • Manifold learning: Isometric Feature Mapping (ISOMAP) Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  10. Methods #2: Graph-Theoretic Clustering • Dominant-Sets clustering • Internal criterion: all objects inside a cluster should be highly similar to each other • External criterion: all objects outside a cluster should be highly dissimilar to the ones inside • Similarity is represented by weights: Dominant-Sets K-means Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  11. Algorithm • Replicator Dynamics approach Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  12. Comparative evaluation • 3 neurons (3 x 300 spikes) • 150 random double-overlaps (3 x 50 spikes) • 50 random triple-overlaps • Variable SNR Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  13. Low SNR example 2 firing neurons (2 x 300 spikes) + + 1 sparse-firing neuron (30 spikes) + overlaps (150 spikes) Neuron 1 Neuron 2 Noise Sparse-firing neuron WaveClus Overlaps Neuron 1 Neuron 2 Sparse-firing neuron Overlaps Graph-Theoretic Spike Sorting Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  14. Conclusions • Problem: • Estimating the number of active neurons • Methods • Methods from the graph-theoretic domain • Replicator dynamics approach • Results • Semi-supervised spike-sorting approach with relative ranking of groups • High ranking: active neurons • Medium ranking: overlapping and noisy spikes that need further processing • Low ranking: noise Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

  15. Thank you Laboratory of Animal Physiology, School of Biology, Aristotle University of Thessaloniki

More Related