ECG data compression by spline approximation

A novel and efficient transform method for ECG data compression based on B-spline basis functions is proposed. The algorithm allows these basis functions to adapt their shape to the nonstationary behavior of ECG signals. The number and shape of these basis functions are completely characterized by the number and location of the so-called knots. The position of the knots can effectively be coded using run-length coding. Therefore, the overhead data required to describe the way the B-spline basis functions vary in time can be neglected compared to the gain in compression achieved by adaptation. The quasi-periodic nature of the ECG signal is used in order to further reduce redundant information in the data. Low bit-rates of the order of 160-200 bits/s are achieved with very good quality of the reconstructed signal. The algorithm is compared with other transform-based schemes (DCT and DLT), and was found superior at any bit-rate.