Filtering beat-to-beat recordings of the high resolution electrocardiogram

The aim of this study was to empirically find a high-pass filter which is suited to high resolution beat-to-beat recordings in that it should not distort the morphology of the QRS complex, while at the same time most closely resemble the magnitude response of the 4 pole Butterworth filter which is usually applied to high resolution signal averaged recordings. The Butterworth filter is applied in a bidirectional mode to the QRS midpoint. Through this, filter ringing does not obscure the endpoints of the averaged QRS, but instead the midportion is affected by artefacts. This represents no real limitation if only the duration and the terminal part of the QRS are considered for analysis. But problems arise if sequences of QRS completes are used to analyse shape changes. The filter was designed and evaluated using a least squares error criterion and three different design procedures. The result was a linear phase FIR filter of order 48 selected among 1248 filters generated.