AcqKnowledge Smoothing

AcqKnowledge software allows you to run a smoothing function as an online calculation or an offline transformation. There are two modes of operation for the smoothing, both use a running average over a user-defined number of samples. In Mean value smoothing, as the moving average is running through the data the mean value is used for the output of the average. In Median value smoothing, as the moving average is running through the data, the median value is used instead of the mean.

The greater the number of samples used for smoothing, the more aggressive the smoothing will be. The tutorial runs through a sample for Mean smoothing and a sample for Median smoothing, plus shows the effect of different values on sample data—and how easy it is to test if you’ve been too aggressive.

The sample file used to demonstrate Mean smoothing is a three channel recording with ECG, Heart Rate, and Respiration; data was recorded at 1,000 samples/second. The heart rate channel is calculated similar to a tachogram, with discreet values and a stair-step display instead of a smooth waveform. Clicking the Transform menu and selecting Smoothing generates the dialog where you select smoothing mode and value. In this sample file, Mean smoothing with 10 sample smoothing has no effect…100 samples begins to take the edges away…500 achieves the desired effect.

For Median smoothing, the sample file has Electrodermal Activity (EDA) recorded at 30 samples/second, a duplicate of the EDA channel to perform the smoothing on, and stimulator firing data. The stimulator was firing pairs of electrical stimuli at a subject and the subject had electrodes on for EDA and the amplifier was picking up the electrical signal as it was delivered from the stimulator. In this type of recording, you’ll want to remove signal contamination without distorting the waveform. Mean smoothing would smooth the data, but Median works very well at removing rapid frequency spikes from a relatively slow-moving signal such as EDA. Median smoothing at 20 samples eliminates the spikes.

To check the signal distortion, hide the stimulator channel, then scale and overlap the two EDA channels….see how noise was eliminated without distorting the waveform. For comparison, switch to Mean smoothing at the same level—you’ll still see noise contamination spikes plus you’ll see distortion of the waveform. Try different modes and different values to identify when Median smoothing is the better choice for some waveforms.

Slew Rate Limiter in AcqKnowledge

This video demonstrates the AcqKnowledge Slew Rate Limiter functionality. The Slew Rate Limiter limits the maximum rate of signal amplitude change over a user defined time period. It can be helpful in removing higher frequency and higher amplitude artifacts from a waveform without distorting the original signal. BIOPAC MRI Solutions.