Concatenation of dictionaries for recovery of ECG signals using compressed sensing techniques

Compressed Sensing (CS) is a promising method for signal recovery using fewer measurement samples than the ordinarily amount imposed by Shannon's sampling theorem. Many fields may need CS technique, but herein we are solely interested in applying it to the physiological signal acquisition systems, particularly the Electrocardiogram (ECG) biosignal. Since this signal is sparse, it is a perfect candidate for CS processing. This paper investigates the application of two greedy algorithms on ECG raw data, namely the matching pursuit (MP) and the orthogonal matching pursuit (OMP) algorithms. Several tests on various ECG data sets are carried out to make the best choice of the parameters (type of dictionary, maximum number of iterations,etc) that allow a good reconstruction of the original signal. Moreover, we propose to concatenate different dictionaries which is shown to enhance the accuracy by 14 dB. The novelty lies in the choice of the dictionaries and the application of the windowing on the original data that allows to significantly reduce the size of the dictionary.