BSL PRO 4 Manual - Biopac

238 BSL PRO ManualThere are, of course, signals that have other frequency signatures in EEG data. Most types of physiologicaldata have a number of different frequency signatures present in the overall signal. In addition, frequencycomponents besides the signal(s) of interest are often present. In the U.S., it is not uncommon for 60 Hzelectrical noise to be acquired along with physiological signals (in other countries, AC interference is presentat either 60 Hz or 50 Hz).Use digital filtering to retain only the frequency components of interest and remove other data (whether it is“noise” or merely physiological signals outside the range of interest).It is important to note that the way in which data is filtered depends in large part on the sampling rate atwhich the original data was acquired. For instance, if data was collected at 50 samples per second (50 Hz), itis not possible to filter out 60 Hz signals.In fact, data must be sampled at a rate equal to at least twice the frequency of the signal to be removed. So, ifdata is to be collected and the information between 80 Hz and 120 Hz is to be removed, the data must besampled at 120 Hz*2, or 240 samples per second (or faster). Also, each channel of data is filtered separately,so removing one type of data from one channel will not affect any other channels.Digital filters can be divided into four general classes:low pass band pass high pass band stopDescriptions of these four classes of filters follow, with visual examples of how these filters work. In each ofthe four examples, a single channel of data containing frequency components in three ranges (low frequency,mid-range, and high frequency) is acquired.• Low frequency data, by definition, has slowly changing values, much like respiration patterns or coretemperature variations.• High frequency data, compared to low frequency data, is noticeably more “spiked,” much like anEMG signal. As you can tell, the high frequency wave repeats itself about five times in the time ittakes the low frequency wave to repeat once.• Mid-range data falls somewhere in between these two extremes.In the examples that follow, one possible way that these data could have been collected is if respiration weremeasured, but the measurement was contaminated with high-frequency muscle movement and mid-frequencysignal coming from AC interference. The data is then passed through a filter, where some of the frequencycomponents are removed.Low pass filteringIn the example below, a low pass filter attenuates the data above a given threshold, allowing only lowerfrequency data to “pass” through the filter.HighfrequencydataMidfrequencydataLowPassFilterLowfrequencydataLowCutoffIncoming dataFiltered dataVisit the online support center at www.biopac.com