A new CS method for ECG signal |
|---|
| Jan Saliga, Pavol Dolinsky, Imrich Andras, Linus Michaeli |
- Abstract:
- Compressed sensing (CS) is a modern method reducing amount of transferred and stored data which have been applied for many types of signals fulfilling sparsity requirements. Electrocardiogram signals (ECG) is one of such signals. In this paper a completely patient-agnostic compressed sensing and reconstruction technique for ECG signals is proposed. A modified sensing method incorporating a QRS detector is used to guarantee the exact R wave positions in the reconstructed signal for any level of compression. For the signal reconstruction a novel method using a dynamic ECG signal model is described, in which the model parameters are found using the Differential Evolution optimization algorithm. Reconstruction quality is evaluated using the MIT-BIH arrhythmia database, and compared against wavelet dictionary reconstruction methods showing better reconstruction quality for compression ratios above 5.
- Download:
- IMEKO-TC4-2020-02.pdf
- DOI:
- -