Spike%20Sorting%20I:

1
Spike Sorting I

• Bijan Pesaran
• New York University

2
Acknowledgements

• Ken Harris and Samar Mehta at Neuroinformatics
  course Woods Hole.

3
Aims

• We would like to
• Monitor the activity of large numbers of neurons
  simultaneously
• Know which neuron fired when
• Know which neuron is of which type
• Estimate our errors

4
THE PROBLEM Multiple Neural Signals
Primate retinal ganglion cells, courtesy of the
lab of Dr. E.J. Chichilnisky
5
THE GOAL Spike Times of Single Neurons
Region from previous slide
Time (sec)
6
THE GRADUATE STUDENT ALGORITHM
7
A GENERAL FRAMEWORK
8
Extracellular Recording Hardware

• You can buy two types of hardware, allowing
• Wide-band continuous recordings
• Filtered, spike-triggered recordings

9
The Tetrode

• Four microwires twisted into a bundle
• Different neurons will have different amplitudes
  on the four wires

10
Raw Data
11
High Pass Filtering

• Local field potential is primarily at low
  frequencies.
• Spikes are at higher frequencies.
• So use a high pass filter. 800hz cutoff is good.

12
Filtered Data
13
Spike Detection

• Locate spikes at times of maximum extracellular
  negativity
• Exact alignment is important is it on peak of
  largest channel or summed channels?

14
Data Reduction

• We now have a waveform for each spike, for each
  channel.
• Still too much information!
• Before assigning individual spikes to cells, we
  must reduce further.

15
Principal Component Analysis

• Create feature vector for each spike.
• Typically takes first 3 PCs for each channel.
• Do you use canonical principal components, or new
  ones for each file?

16
Feature Space
17
Cluster Cutting

• Which spikes belong to which neuron?
• Assume a single cluster of spikes in feature
  space corresponds to a single cell
• Automatic or manual clustering?

18
Cluster Cutting Methods

• Purely manual time consuming, leads to high
  error rates.
• Purely automatic untrustworthy.
• Hybrid less time consuming, lowest error rates.

19
Semi-automatic Clustering
20
How Do You Know It Works?

• We can split waveforms into clusters, but are we
  sure they correspond to single cells?
• Simultaneous intra- and extra-cellular recordings
  allow us to estimate errors.
• Quality measures allow us to guess errors even
  without simultaneous intracellular recording.

21
Intra-extra Recording

• Simultaneous recording with a wire tetrode and
  glass micropipette.

22
Intra-extra Recording
Extracellular waveform is almost minus derivative
of intracellular
23
Bizarre Extracellular Waveshapes
Model
Experiment
24
Two Types of Error

• Type I error (false positive)
• Incorrect inclusion of noise, or spikes of other
  cells
• Type II error (false negative)
• Omission of true spikes from cluster
• Which is worse? Depends on application

25
Manual Clustering Contest
26
Best Ellipsoid Error Rates
Find ellipsoid that minimizes weighted sum of
Type I and Type II errors. Must evaluate using
cross-validation!
27
Humans vs. B.E.E.R.
28
Waveshape Helps Separation
29
Why were human errors higher?

• To understand this, try to understand why
  clusters have the shape they do
• Simplest possibility spike waveform is constant,
  cluster spread comes from background noise
• Are clusters multivariate normal?

30
Problem Overlapping Spikes
31
Problem Cellular Synchrony
32
Problem Bursting
33
Problem Misalignment

• When you have a spike whose peak occurs at
  different times on different channels, it can
  align on either.
• This causes the cluster to be split in two.

34
Problem Dimensionality
Manual clustering only uses 2 dimensions at a
time BEER measure can use all of them
35
Semi-Automatic Clustering

• Uses all dimensions at once
• Errors should be lower
• Still requires human input

36
Semi-automatic Performance
37
Software KlustaKwik
klustakwik.sourceforge.net

• Mixture of Gaussians, unconstrained covariance
  matrices
• Speed is crucial
• CEM Algorithm faster than EM
• Most probabilities not calculated
• Local maxima result in over- and under-clustering
• Split and merge features to tunnel out of local
  maxima
• Still requires supercomputer resources.

38
Software Klusters
klusters.sourceforge.net
Recluster Feature
Ergonomic Design
39
Cluster Quality Measures

• Would like to automatically detect which cells
  are well isolated.
• BEER measure needs intracellular data, which we
  dont have in general.
• Will define two measures that only use
  extracellular data.

40
Isolation Distance
Size of ellipsoid within which as many spikes
belong to our cluster as not
41
L_ratio
42
False Positives and Negatives
43
Which Measure to Use?

• Isolation distance correlates with false positive
  error rates
• Measures distance to other clusters
• L_ratio correlates with false negative error
  rates
• Measures number of spikes near cluster boundary

44
Conclusions

• Automatic clustering will save time and reduce
  errors.
• Errors can be as low as 5.
• Quality measures give you a feeling of how bad
  your errors are.

45
Room for Improvement
Easy

• Make it faster
• Improved spike detection and alignment
• Quality measures that estimate error
• Fully automatic sorting
• Resolve overlapping spikes

Hard