Performance of RLS and LMS algorithms in KL estimation of ischemic ECG records

The Karhunen-Loeve (KL) transform is a tool to analyse the repolarization period in the ECG. Adaptive algorithms improve the KL series estimation. The recursive least squares (RLS) and least mean squares (LMS) algorithms are studied when applied to estimate the KL coefficients of the ST-T complex in the ECG signal. The performance of RLS and LMS algorithms are compared both in improvement of signal-to-noise ratio (SNR) and in convergence rate. It is presented a specific initialization for the LMS algorithm that obtains the same performance than RLS with lower calculations and without the numerical instability problem, making it the most suitable for the KL estimation.