Event Tab

NeuroMatic analysis software for detecting spontaneous events. This page describes the threshold-below-baseline (or threshold-above-baseline) detector similar to that described by Kudoh and Taguchi 2002.

Threshold < Baseline. A forward sliding-window search for the first data point that falls below level y at dt msec after t₀ (negative events, see image below) or falls above level y at dt msec after t₀ (positive events, threshold > baseline). This algorithm is similar to that of Kudoh and Taguchi 2002 except for the use of a baseline average (avg) window centered at t₀ rather than a single baseline point at t₀.

avg = baseline average window size

Y_avg = baseline average y-value

y = Y_avg - threshold (threshold < baseline, negative events)

y = Y_avg + threshold (threshold > baseline, positive events)

t₀ = baseline window midway-time-point (slides forward during event detection)

dt = event detection window beginning at t₀

t_event = y-threshold-crossing event time (t₀ + dt)

There are two baseline parameters, avg and dt. Baseline avg specifies the baseline average window size (orange line). Baseline dt specifies the time after the baseline midway-time-point to search for an event; this parameter is equivalent to wi of Kudoh and Taguchi 2002. Previously avg and dt were set equal, but now you have the option to specify a larger dt without increasing the baseline window. If you do not wish to use a baseline window, set avg = 0.

Event Onset Search. This is an optional search creteria for the onset of an event once the event has been detected. If no onset is found, the event is ignored and the search for events continues. If you do not wish to use this onset criteria for limiting what is considered an acceptable event, uncheck the onset checkbox.

The onset search is a backward sliding-window search from t_event for the first occurrence of the right-most window point falling above level y (negative events, see image below) or below level y (positive events). Note, this algorithm is slightly different from Kudoh and Taguchi 2002 in that it searches backward from t_event. Searching backward gives a more accurate estimate of the onset time since it is less influenced by the baseline noise. There are three onset search parameters which you can adjust manually: avg, Nstdv and limit.

avg = sliding average window size (slides backward from t_event)

Y_avg = average y-value computed within avg window

Y_stdv = standard deviation of y-values computed within avg window

Nstdv = number of standard deviations above / below Y_avg

y = Y_avg - Nstdv * Y_stdv (negative events)

y = Y_avg + Nstdv * Y_stdv (positive events)

t_onset = y-threshold-crossing time

limit = time before t_event to end onset search

Event Peak Search. This is an optional search creteria for the peak of an event once the event has been detected. If no peak is found, the event is ignored and the search for events continues. If you do not wish to use this peak criteria for limiting what is considered an acceptable event, uncheck the peak checkbox.

The peak search is a forward sliding-window search from t_event for the first occurrence of the left-most window point falling below level y (negative events, see image below) or above level y (positive events). Note, this algorithm implements only the forward peak search of Kudoh and Taguchi 2002. There are three peak search parameters: avg, Nstdv and limit.

avg = sliding average window size (slides forward from t_event)

Y_avg = average y-value computed within avg window

Y_stdv = standard deviation of y-values computed within avg window

Nstdv = number of standard deviations above / below Y_avg

y = Y_avg - Nstdv * Y_stdv (negative events)

y = Y_avg + Nstdv * Y_stdv (positive events)

t_peak = y-threshold-crossing time

limit = time after t_event to end peak search

Equivalent Parameters of Kudoh and Taguchi 2002 for Threshold Search

wi = dt = sliding detection window (they use 1.6 - 4 msec)

bin = onset avg window = peak avg window (they use 1 - 2 msec)

pp = peak limit (they use 4 - 10 msec)

n = onset Nstdv = peak Nstdv (they use 1 - 1.5)

baseline avg they do not use, in which case avg = 0 msec.

Adjusting Parameters

• Set threshold/level to minimum event amplitude to be detected.

• Set sliding detection window dt such that the midway-time-point of the baseline average window t₀ rests comfortably before event time t_event. The onset limit value can be set to this same value.

• Set peak search limit to approximately half the length of your event.

• Set onset and peak Nstdv to 1 - 2, depending on the noise levels in your data.

• Adjust peak, onset and baseline avg windows to obtain best results. These values will change depending on the levels of noise in your data.