Noise processing in exercise ECG stress test for the analysis and the clinical characterization of QRS and T wave alternans

The aim of this study is to analyze different sources of noise in ECG recordings from stress tests in order to obtain reliable parameters and measurements for the analysis of T-wave and QRS-complex alternans (TWA & QRSA). Simple methods for eliminating common sources of noise like power-line interference, baseline drift and electromyographic noise were used. The pre-processing phase considered the detection of steep slope/spike, low amplitude signal, and flat line or missing lead artefacts. The detection of TWA and QRSA was based on Principal Component Analysis indices and wave amplitudes considering all the leads. A particular database of 106 ECG records during stress testing was considered. The signal quality analysis performed in this study has permitted to obtain reliable and noise-tolerant measurements of TWA and QRSA indices. The different diagnostic groups were used for the evaluation of the clinical significance of the alternans. Men have significantly higher values of QRSA than women. Smokers have significantly higher TWA values as compared with non-smokers. Significant negative correlation was obtained between age and both TWA and QRSA. Correlations between TWA and QRSA and the double product of arterial hypertension and the maximal heart rate during the stress test were statistically significant, positive and relatively strong.