Other facilities

DataView has several general-purpose analysis facilities, including

• Phase-Plane Analysis
  In phase-plane analysis, one time-dependent variable is plotted against another. A common use of phase-plane plots in neuroscience is to provide an alternate visualization of the shapes of features such as spikes or PSPs, which is achieved by plotting the membrane potential (V) against its time derivative (dV/dt) at each moment in time over the duration of the feature. The phase-plane plot can reveal subtle changes in shape over time that are difficult to pick up in an extended record.

The left panel shows a whole-cell patch recording from a spinal motorneurone in the tadpole. The right panel shows a 3-D phase-plane display of the record, coloured with respect to time. Note the cycle between the yellow and green stages, where the spike failed.

 

• Spike Shape Analysis
  Spike threshold can be detected either as an absolute voltage threshold, or the voltage at which a user-specified rate of chang of voltage occurs, or the voltage at which the rate of change of voltage changes most rapidly. From the threshold various metrics defining spike shape are calculated and can be displayed.

• Lorentz (Poincaré) Return Map
  This technique simply involves plotting the value of an item in a sequence against the value of the next item in the sequence. Typically, each item could be the peak (or trough) value of successive cycles in an oscillating system. Plots of this sort can reveal “meta”-rhythms, i.e. rhythms within rhythms like the beat frequency of mixed oscillators, and is used in investigating biological phenomena such as heart beat and respiratory or locomotory rhythms.

The left panel shows a noisy sine wave with a 25 ms cycle period. A regular event train have been generated with a 10 ms interval. This means that the events will return to approximately the same value in the data trace every 5th event (50 ms). So there will be 5 different values of Vn/Vn+1 associated with the events. These values are plotted in the right-hand panel, which is a Lorentz return map. Note that because the sine wave is noisy, the values form clusters, rather than exactly-superimposing points. Again, the colour represents the time in the data trace for each point in the plot.

• Waveform correlation
  Auto- and cross-correlation analysis can be performed on both waveform and event data.

The left panel shows a waveform that develops a strong oscillation pattern (it's actually a fly singing). The right panel shows the autocorrelation of the left panel, up to a lag of 10 ms.

• Cumulative sign-correlation detection of common features
  This technique helps detect waveform transients (such as PSPs) which occur simultaneously in two channels of data.

The top trace shows a neuron that makes excitatory input to the neurons in the 2nd and 3rd traces (data simulated using Neurosim). Sampling at 50 ms intervals throughout the record shows 40 same-sign intances between these traces, and 12 opposite-sign instances. This gives a cumulative sign correlation of 0.4 (bottom trace). The binomial probability of this, given no common signal (equal likelihood of same and opposite sign instances), is less than 0.01.

• Joint Peristimulus Time Histogram (JPSTH) )
  A JPSTH can be constructed as descibed by Aertsen et al. (1989, J. Neurophysiol. 61; 900-917), and the MU lab website.

• Stimulus-Response Latency Raster
  Raster plots can reveal the effects of a stimulus on spike frequency.
  
  
  

• Audification (a.k.a. Sound)
  Any data file can be played as a sound. You can either play the data at the recording frequency, or, if this is too low, you can select 5 or 11 kHz as the sound frequency. Listening to data played as sound can be a good way of detecting subtle changes in frequency or pattern within the data.
  This facility utilises the DirectX sound capabiliy of PCs.