Link

Standard cell metrics

CellExplorer used a single Matlab struct for handling all cell metrics called cell_metrics. The cell_metrics struct consists of three types of fields for handling different types of data: double, char cells and structs. Fields must be defined for all cells in the session (1xnCells). Single numeric values are saved to numeric fields with double precision, and character/string fields are saved in char cell arrays. Time series data like waveforms and session parameters are stored in standard struct fields.

Table of contents

  1. General metrics
  2. Spike events based metrics
  3. ACG & CCG based metrics
  4. Waveform based metrics
  5. Cell-type classification
  6. Monosynaptic connections
  7. Sorting quality metrics
  8. Sharp wave ripple metrics
  9. Theta oscillation metrics
  10. Firing rate maps
  11. Spatial metrics
  12. Firing rate stability metrics
  13. Event metrics
  14. Manipulation metrics
  15. Response curve metrics
  16. Group data

General metrics

  • general : struct
    • basename : name of the session
    • basepath : full path to the session
    • cellCount : number of cells in the session
    • processinginfo. The general fields also contains a list of timestamps for time-series metrics.
  • brainRegion: Brain region acronyms from Allan institute Brain atlas.
  • animal: Unique name of animal.
  • sex: Sex of the animal [Male, Female, Unknown]
  • species: Animal species [Rat, Mouse,…]
  • strain: Animal strain [Long Evans, C57B1/6,…]
  • geneticLine: Genetic line of the animal
  • spikeGroup: Spike group: Shank number / spike group.
  • labels: Custom labels.

Spike events based metrics

  • spikes: struct containing spike times
    • times : spike times in seconds for each cell (stored as a cell array following the format of the struct spike.times).
  • spikeCount: Spike count of each cell from the entire session (numeric).
  • firingRate: Firing rate in Hz: Spike count normalized by the interval between the first and the last spike.
  • cv2: Coefficient of variation (CV_2).
  • burstIndex_Mizuseki2012 Burst index: Fraction of spikes with a neighboring ISI < 6 ms as defined by Mizuseki et al. Hippocampus 2012.

ACG & CCG based metrics

  • acg: autocorrelograms. Three types:
    • wide [-1000 ms : 1 ms: 1000 ms]
    • narrow [-50 ms : 0.5 ms : 50 ms]
    • log10 [log-intervals spanning 1 ms : 10 s].
  • isi: interspike intervals
    • log10 [log-intervals spanning 1 ms : 10 s].
  • Autocorrelograms are fitted with a triple-exponential equation:
    ACG_fit = 'max(c*(exp(-(x-f)/a)-d*exp(-(x-f)/b))+h*exp(-(x-f)/g)+e,0)'
    a = tau_decay, b = tau_rise, c = decay_amplitude, d = rise_amplitude, e = asymptote, f = refrac, g = tau_burst, h = burst_amplitude
    
\[ACG_{fit} = max(c\exp(\frac{-(x-t_{refrac})}{\tau_{decay}})-d\exp(\frac{-(x-t_{refrac})}{\tau_{rise}})+h\exp(\frac{-(x-t_{refrac})}{\tau_{burst}})+rate_{asymptote},0)\]

See the dedicated page about the fitting procedure.

  • acg_tau_rise ACG tau rise (ms)
  • acg_tau_decay ACG tau decay (ms)
  • acg_tau_burst ACG tau bursts (ms)
  • acg_refrac ACG refractory period (ms)
  • acg_fit_rsquare ACG fit R-square
  • thetaModulationIndex is defined by the difference between the theta modulation trough (mean of autocorrelogram bins 50-70 ms) and the theta modulation peak (mean of autocorrelogram bins 100-140ms) over their sum.
  • synapticEffect: Synaptic effect
  • burstIndex_Royer2012 Burst index (Royer 2012)
  • burstIndex_Doublets Burst index doublets.
  • synapticConnectionsIn: Synaptic ingoing connections count.
  • synapticConnectionsOut: Synaptic outgoing connections count.

Waveform based metrics

  • waveforms: spike waveform struct with below fields:
    • filt: Average filtered spike waveform from channel with max amplitude. High-pass filtered above 500Hz to standardize waveforms.
    • raw: Average raw spike waveform from channel with max amplitude.
    • time: Time vector for average raw spike waveform from channel with max amplitude.
  • maxWaveformCh: Max channel zero-indexed: The channel where the spike has the largest amplitude.
  • maxWaveformCh1: Max channel one-indexed: The channel where the spike has the largest amplitude.
  • troughToPeak: Trough-to-peak latency is defined from the trough to the following peak of the waveform.
  • ab_ratio: Waveform asymmetry; the ratio between the two positive peaks (peakB-peakA)/(peakA+peakB).
  • peakVoltage: Peak voltage (µV) Defined from the channel with the maximum waveform (high-pass filtered). max(waveform)-min(waveform).

Cell-type classification

Monosynaptic connections

  • putativeConnections: putative connections determined from cross correlograms. Contains two fields: excitatory and inhibitory, each contains connections pairs.

Sorting quality metrics

Isolation distance and L-ratio as defined by Schmitzer-Torbert et al. Neuroscience. 2005.

  • isolationDistance: Isolation distance.
  • lRatio: L-ratio.
  • refractoryPeriodViolation: Refractory period violation (‰): Fraction of ISIs less than 2 ms.

Sharp wave ripple metrics

  • ripples_modulationIndex: strength of ripple modulation of the firing rate)
  • ripples_modulationPeakResponseTime: Ripple peak delay. Calculated from a ripple triggered average. The delay between the ripple peak and the peak response of the ripple triggered average response.
  • deepSuperficial: Deep-Superficial region assignment [Unknown, Cortical, Superficial, Deep].
  • deepSuperficialDistance: Deep Superficial depth relative to the reversal of the sharp wave. (in um).

Theta oscillation metrics

  • thetaPhasePeak: Theta phase peak
  • thetaPhaseTrough: Theta phase trough
  • thetaEntrainment: Theta entrainment
  • thetaModulationIndex: Theta modulation index. Originally defined in Cacucci et al., JNeuro 2004. Computed as the difference between the theta modulation trough (defined as mean of autocorrelogram bins, 50-70 msec) and the theta modulation peak (mean of autocorrelogram bins, 100-140 msec) over their sum.

Firing rate maps

  • firingRateMaps: (spatial) firing rate maps.

Spatial metrics

The spatial metrics are all based on average firing rate map.

  • spatialCoverageIndex: Spatial coverage index. Defined from the inverse cumulative distribution, where bins are sorted by decreasing rate. The 75 percentile point defines the spatial coverage by the fraction of bins below and above the point (defined by Royer et al., NN 2012)
  • spatialGiniCoeff: Spatial Gini coefficient. Defined as the Gini coefficient of the firing rate map.
  • spatialCoherence: Spatial Coherence. Defined by the degree of correlation between the firing rate map and a hollow convolution with the same map (reference?)
  • spatialPeakRate: Spatial peak firing rate (Hz). Defined as the peak rate from the firing rate map.
  • placeFieldsCount: Number of place fields. Defined as the number of intervals along the firing rate map that fulfills a number of spatial criteria: minimum rate of 2Hz and above 10% of the maximum firing rate bin and minimum of 4 connecting bins. The cell further has to have a spatial coherence greater than 0.6 (Mizuseki et al ?).
  • placeCell: Place cell (binary, determined from the Mizuseki spatial metrics).

Firing rate stability metrics

  • firingRateGiniCoeff: The Gini coefficient of the firing rate across time.
  • firingRateStd: Standard deviation of the “firing rate across time” divided by the mean.
  • firingRateInstability: Mean of the absolute differential “firing rate across time” divided by the mean: abs(diff(firingRateAcrossTime)).

Event metrics

  • events: event time series.

Manipulation metrics

  • manipulations: manipulations time series.

Response curve metrics

  • responseCurves: response curves.

Group data

  • groups: Cell groups. Each cell can be assigned to one or more groups.
  • tags: Each cell can be assigned to one or more tags.
  • groundTruthClassification: Opto-tagged/ground truth cell groups. Each cell can be assigned to one or more groups.