Sampling Theorem and Aliasing in Biomedical Signal Processing

Abstract Despite digital techniques for data acquisition and processing being widely used in biomedical research for quite some time, inappropriate signal conditioning and digitization are still potential pitfalls threatening both the reliability of the experiments and the proper interpretation of the acquired data. The aim of this chapter is to review: (1) the Sampling (Nyquist) Theorem; (2) the concept of aliasing; (3) the importance of antialiasing low‐pass filtering for eliminating the effect of aliasing and appropriately determining the sampling frequency; (4) the advantages of properly chosen filter cut‐off frequency and slope for determining the minimal required sampling frequency; and (5) the impact of incorrectly selected sampling frequency on the interpretation of biomedical data. In a case study, a model of electrogastrographic (EGG) recording is mixed with a model of electrocardiographic (EKG) artifact in an overall white noise environment. The resulting composite signal is low‐pass filtered and then digitized with a sampling frequency of 1 Hz. The cut‐off frequency of the first‐order low‐pass filter is altered from 0.5 Hz to 0.1 Hz. Amplitude frequency spectra of the digitized recordings are investigated to illustrate the effect of aliasing. An example with a real human electrogastrogram, in which an EKG artifact is present, illustrates the simulation results. When a first‐order antialiasing filter is used, at least a five‐fold difference between the filter cut‐off frequency and the sampling frequency is recommended for compliance with the Sampling Theorem. Increasing the order of the antialiasing filter can reduce the required sampling frequency, but can also make the entire instrumentation system underdamped, thus injecting oscillatory artifacts every time abrupt or sudden changes in the external conditions during the recording occur.