Periodic Noise Filter for Electrophysiological Signals with Interfering Transients

Periodic noise such as power line noise is always a serious problem in measurements of complex temporal signals, for instance EEG, ECG and MEG signals in both neuroscience research and clinical environment. Many methods have been developed to remove period noise. However, none of them works well if one or more types of transient interference exist in the recordings. Transient interferences such as stimulus artifact are defined as those transient signals that contain energies in the frequency range of the period noise to be filtered out. When there is an interfering transient, the traditional methods can not remove the periodic noise without distortion of the signal interested. Based on the Single Tone Extraction Method (STEM), which is a well developed algorithm for estimating the frequency, amplitude and phase of a periodic signal or single tone in complex temporal signals, we have developed a new versatile, useful method to remove the periodic noise such as power line noise when the data to be analyzed contain transient interference(s).

This software and  method, called RAW-STEM, can be applied to complex temporal signals with one or more interference transients.

Our technology is a significant improvement over traditional filtering methods and the STEM for several reasons. First, it does not suffer from the problem of ringing following the stimulus artifact regardless of size of the artifact, which distorts the signals. This technology does not introduce spikes in the recordings as it is the case for comb filters when the stimulus artifact is strong. Second, it is fast enough to be used for on-line analyses of temporal signals. The method requires a few milliseconds to remove the noise using ordinary PC computers. Third, the method estimates the parameters of the periodic noise with high accuracy within the range of signal to noise that is commonly seen in electrophysiological recordings. The accuracy is 0.007-2%, depending on the noise level and the periodic noise parameter. Fourth, it can be applied even when there are multiple interfering transients in the recording. It can be readily embedded into a data acquisition system, or built into a custom DSP chip. Although this method was proposed and its utility has been demonstrated for electrophysiological data, it can be widely used in a wide variety of measurements of complex temporal signals with interfering transient(s) and control system applications.

Possible applications:

1. Electrophysiological signal recording in both basic research and clinical applicaiton
2. Biomagnetic studies in both basic and clinical research;
3. MRI research with high-frequency unwanted periodic noise and transient interferences such as gradient transients.
4. Control system applications involving electric or electronic signals as signals to be controlled.
5. Radio-communications in which data are contaminated by unwanted periodic noise and interfering transients.

This product is a plug-in that is used with LabVIEW.