Python based bio-signal processing: Mitigation of baseline wandering in pre-recorded electrooculogram

Baseline wandering is a common artefact observed in bio-signals, particularly in electrocardiogram (ECG), electrooculogram (EOG), and electroencephalogram (EEG) signals. It refers to the slow drift or fluctuation of the baseline of the signal, which can obscure the underlying physiological information. Baseline wandering can be caused by various factors, which introduce low-frequency noise into the signal, resulting in a shifting baseline. One of the biggest challenges in bio-signal processing is mitigating baseline wandering without compromising the signal quality. In this chapter, we introduce baseline wandering and its origin, using a pre-recorded EOG signal as an example. We discuss techniques such as the normalisation and standardisation of an EOG, wavelet decomposition of EOG signals, and estimations of signal power at different levels. We also demonstrate the filtering of the baseline drift on EOG samples using a threshold set based on wavelet energy. The chapter is complemented with a hands-on Python code that covers loading the signal, wavelet decomposition, and baseline correction.