Comparison of four methods of averaging nerve activity.

Four methods of averaging nerve activity, moving time average (Analog), integration (Integrated), counting spikes (Spikes), and counting pulses from a voltage-to-frequency converter (VFC), were used to analyze artificial pulse trains and renal, carotid sinus, and vagal nerve activities. Results of the methods were compared using least-squares linear regression and correlation to determine the linearity of each method with respect to changes in frequency, amplitude, and width of pulse trains and the degree of agreement between methods. The methods that respond to total voltage (Analog, Integrated, and VFC) were linear with respect to input pulse train modulations and agreed closely with each other when averaging pulse trains, summating pulses, and nerve activity. Spikes were linear with respect to frequency modulation but not with respect to amplitude changes, pulse width changes, or pulse summations. In general, Spikes did not agree as well with Analog, Integrated, and VFC as these methods agreed with each other when averaging nerve activity. The degree of agreement was a function of the voltage threshold for Spikes and the level of nerve activity. Two methods of minimizing noise and obtaining a zero reference level for nerve activity were compared: setting a voltage threshold, such that noise was below and activity above threshold, was found to shift the base-line activity toward zero and compress phasic changes in activity; and recording the average noise level from a crushed nerve and subtracting it from averaged activity shifted the base-line activity toward zero with no change in the phasic component.