CellExplorer used a single Matlab struct for handling all cell metrics called
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.
- General metrics
- Spike events based metrics
- ACG & CCG based metrics
- Waveform based metrics
- Cell-type classification
- Monosynaptic connections
- Sorting quality metrics
- Sharp wave ripple metrics
- Theta oscillation metrics
- Firing rate maps
- Spatial metrics
- Firing rate stability metrics
- Event metrics
- Manipulation metrics
- Response curve metrics
- Group data
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.
spikes: struct containing spike times
times: spike times in seconds for each cell (stored as a cell array following the format of the struct
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_Mizuseki2012Burst index: Fraction of spikes with a neighboring ISI < 6 ms as defined by Mizuseki et al. Hippocampus 2012.
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_tau_riseACG tau rise (ms)
acg_tau_decayACG tau decay (ms)
acg_tau_burstACG tau bursts (ms)
acg_refracACG refractory period (ms)
acg_fit_rsquareACG fit R-square
thetaModulationIndexis 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_Royer2012Burst index (Royer 2012)
burstIndex_DoubletsBurst index doublets.
synapticConnectionsIn: Synaptic ingoing connections count.
synapticConnectionsOut: Synaptic outgoing connections count.
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
peakVoltage: Peak voltage (µV) Defined from the channel with the maximum waveform (high-pass filtered).
putativeCellType: Putative cell types.
- In the processing pipeline, cells are classified into three putative cell types: Narrow Interneurons, Wide Interneurons and Pyramidal Cells.
- Interneurons are selected by 3 separate criteria:
- acg_tau_decay > 30 ms
- acg_tau_rise > 3 ms
- troughToPeak <= 0.425 ms
- Next interneurons are separated into two classes
- Narrow interneuron assigned if troughToPeak <= 0.425 ms
- Wide interneuron assigned if troughToPeak > 0.425 ms
- Remaining cells are assigned as Pyramidal cells.
putativeConnections: putative connections determined from cross correlograms. Contains two fields:
inhibitory, each contains connections pairs.
Isolation distance and L-ratio as defined by Schmitzer-Torbert et al. Neuroscience. 2005.
isolationDistance: Isolation distance.
refractoryPeriodViolation: Refractory period violation (‰): Fraction of ISIs less than 2 ms.
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).
thetaPhasePeak: Theta phase peak
thetaPhaseTrough: Theta phase trough
thetaEntrainment: Theta entrainment
thetaModulationIndex: Theta modulation index. determined from the ACG.
firingRateMaps: (spatial) firing rate maps.
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).
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:
events: event time series.
manipulations: manipulations time series.
responseCurves: response curves.
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.