Name
Abstract | ||
|---|---|---|
We present a novel approach to single-channel ECG-EMG signal separation by means of enhanced non-negative matrix factorization (NMF). The approach is based on a linear decomposition of the input signal spectrogram in two non-negative components, which represent the ECG and EMG spectrogram estimates. As ECG and EMG have different time-frequency (TF) patterns, the decomposition is enhanced by reshaping the input mixture spectrogram in order to emphasize a sparse ECG over a noisy-like EMG. Moreover, initialization of the classical NMF algorithm with accurately designed ECG and EMG structures further increases its separation performance. The comparative study suggests that the proposed method outperforms two reference methods for both synthetic and real signal mixture scenarios. |
Year | DOI | Venue |
|---|---|---|
2014 | 10.1109/EMBC.2014.6944553 | EMBC |
Keywords | Field | DocType |
electrocardiography,time-frequency patterns,input mixture spectrogram,ecg spectrogram estimates,emg spectrogram estimates,enhanced nonnegative matrix factorization,medical signal processing,nonnegative components,source separation,input signal spectrogram,matrix decomposition,electromyography,ecg-emg signal separation,linear decomposition,classical nmf algorithm,time-frequency analysis | Pattern recognition,Computer science,Electronic engineering,Speech recognition,Non-negative matrix factorization,Artificial intelligence | Conference |
Volume | ISSN | Citations |
2014 | 1557-170X | 1 |
PageRank | References | Authors |
0.36 | 4 | 2 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Maciej Niegowski | 1 | 1 | 1.37 |
| Zivanovic, M. | 2 | 59 | 10.80 |