A recursive algorithm to compute the baseline drift in recorded biological signals

Baseline estimation and removal is one of the main problems to overcome in order to obtain a correct interpretation of recorded biological signals. The method described in this paper is based on a new application of recursive estimation of a smoothing spline, selected to model the unknown baseline. Its advantages over conventional methods derive from these characteristics: it is parametric and recursive, it works in the time domain, and the same software can be used in different applications, since no a priori frequency knowledge is needed. In the example, the method is applied to ECG recordings and a spectral analysis is thereby shown.