A novel algorithm for cardiac biosignal filtering based on filtered residue method.

In this paper, a new algorithm is presented for the filtering (de-noising) of cardiac bioelectrical signals. The primary target of this algorithm is the class of cardiac action potentials recorded using voltage-sensitive dyes, although the method is also applied to electrocardiographic signals. High periodicity is one of the main features of cardiac biosignals. The proposed algorithm exploits this feature in filtering signals with a minimum amount of distortion. The basic idea is to use signal averaging in time to find the stationary portion of the signal. The residue is found by subtracting the signal average from the corresponding points of the input. After passing through a low-pass filter, the filtered residue (FR) is added back to the signal average to reconstruct the output. The practical implementation of the filter residue algorithm is discussed. Stretching and shrinking operations are the basis for the conversion of quasi-periodic signals into periodic signals, which can then be subjected to the FR algorithm. Various examples are presented, and error estimation is performed to guide the selection of optimal parameters for the algorithm. The ability of the algorithm to reconstruct the variation among beats is demonstrated, and its limitations are discussed.