Multi-scale dictionary learning for compressive sensing ECG

Compressed sensing (CS) as applied to the electrocardiogram (ECG) utilizes the sparsity of ECG signals to enable accurate reconstruction from undersampled data. Most prior work in compressive sensing ECG has employed analytical sparsifying transforms such as wavelets. In this paper, we propose to adaptively learn a sparsifying transform (dictionary) that exploits the multi-scale sparse representation of ECG signals. By calculating subdictionaries at different data scales, we are able to exploit the correlation within each wavelet subband and, subsequently, represent the data in a more efficient manner. Numerical experiments, conducted on records selected from the MIT-BIH arrhythmia database, demonstrate the good performance of the proposed method in terms of reconstruction error.