Comparison of four methods of averaging nerve activity

Francis A. Hopp, Jeanne L. Seagard, John P. Kampine 

Am J Physiol (1986) 

Changes in nerve activity is used to monitor the response of a physiological system. Measuring nerve activity has been accomplished using many different methods, depending on what is being researched. Some methods focus on how nerve activity is related to total number of active fibers or some relate nerve activity to total voltage. At the time of this paper it was not determined if any method of analysis was better than others. The most common are counting spikes (Spikes), measuring the moving time average (analog), integrating the neural signal (Integrated), and processing activity using a voltage-to-frequency converter (VFC). This paper compares characteristics of these four averaging techniques.

Mongrel dogs were anesthetized and maintained using a fusion pump. The nerve activity was recorded from carotid sinus nerve, the cervical vagus, and a renal nerve. The data was analyzed by processing the activity using the four types of averaging systems. Pulse trains were used to verify the linearity of each averaging method with respect to changes in frequency, amplitude, and width. Analysis was also done by performing regressions on the average outputs of each method against the other methods for the same input.

When measuring the moving time average, full wave rectified nerve activity was processed using a low-pass filter that smoothed spikes into a time-varying voltage level with an instantaneous amplitude proportional to the total voltage of the electroneurogram per averaging interval. In the VFC averaging method, nerve activity was full wave rectified and processed through a VFC with a gain of 10kHz/V. The output frequency of the VFC was proportional to the instantaneous value of the total voltage of the electroneurogram. VFC pulses were then counted in 100 ms intervals which generated a stair-step voltage proportional to the average activity for the previous interval. Using the integration method, nerve activity was full wave rectified and integrated. The integrator was reset ever 100 ms. The result was a signal processed using a peak detector to generate a stair-step proportional to the total integrated nerve activity per unit time for the previous interval. In the counting spikes method, a multiplexing circuit was used to display both the neural signal voltage threshold on a single oscilloscope channel. The threshold was set by eye to a level just above the noise. Each neural spike above the set threshold was then counted.

The study showed that all averaging methods are linear with respect to frequency modulated pulse trains. Analog, integrated , and VFC are linear with respect to width and amplitude, but spikes responses in a nonlinear way to amplitude modulation and does not respond to pulse width modulation. Analog, integrated, and VFC are able to average pulse summations, while spike output in reduced. This shows that counting spikes should be used for a preparation with fewer nerve fibers. 

Although this paper is from 1986, analyzing nerve activity today is not extremely different. The ability to pick out individual axon’s activity from raw bulk records has still evaded researchers, but science is getting very close. This study of four methods helps me further understand the process I am currently working on in lab by showing limitations of older methods and by forcing the reader to think of how new methods must be designed to sharpen the results.

-Paul M     (My childhood nickname was Bambi)