Alert and Surveillance System for Newborns

A Study on Dictionary Selection in Compressive Sensing for ECG Signals Compression and Classification

“…Let us stress the fact that although pure sparse signals (built of exactly k<<N atoms from a specified dictionary) are difficult to find, conventional results are valid for signals that are "almost sparse" (which can be built of k<<N non-negligible atoms) with respect to dictionaries that can be overcomplete (contain more atoms than their intrinsic dimension), as in the case of some classes of biomedical signals. Taking into consideration this fact, it has been found useful to adapt the theory of CS to the field of processing ECG and electroencephalographic (EEG) signals [2][3][4] as well as for applications [5] such as compression, transmission, reconstruction of ECG signals, ECG filtering and monitoring [6,27,[30][31][32].…”

“…Many of the papers that address CS focus on how to build specific dictionaries for signal reconstruction [13][14][15][16][17][18][19][20][21][22][23][24][25][26]. In the case of the ECG signal, due to its particularities, namely, the quasi-periodicity of the P, Q, R and S waves and the preservation of their shapes, many of the methods proposed in the literature focus on the advantages offered by these features specific to the ECG signal [27][28][29][30][31][32][33][34][35][36][37]. Thus, a large part of the methods proposed regarding CS of ECG signals aim at building dictionaries specific to these signals.…”