Interpolation technique for extracting features from ECG signals sampled at low sampling rates

The occurrence of the R peak (t_R), and the area of the QRS complex (A_QRS) are among the features estimated from each ECG wave. To estimate these parameters ECG is traditionally sampled at a high sampling rate (250-500 Hz). However, since t_R and A_QRS are identified by the selective analysis of the QRS complex and since the QRS maximal frequency component is about 25 Hz, we propose a new procedure to identify t_R and A_QRS from ECG sampled at the relatively low rate dictated by the frequency content of the QRS complex. The procedure identifies each QRS complex from the sampled ECG, and estimates t_R and A_QRS after interpolation of the QRS complex. The interpolation is obtained as the inverse discrete Fourier transform (IDFT) of the zero-padded DFT of the QRS. The technique was tested on a public database of ECG. Results showed that by this technique t_R and A_QRS can be accurately derived from ECG sampled at frequency as low as 70 Hz.