A new mathematical based QRS detector using continuous wavelet transform

In this paper, a new viewpoint in ECG detection is presented using continuous wavelet transform (CWT). In order to magnify QRS complex and reduce the effects of other peaks, the concept of dominant rescaled wavelet coefficients (DRWC) is defined. Using this concept, the relations between the time duration of components of a QRS complex and their wavelet transforms are derived analytically. The proposed relations are used to define local search interval at the vicinity of each QRS complex components. Using DRWC concept, the proposed detection algorithm enables us to detect the R peaks even at the presence of long P and T peaks. Then, each detected complex is classified based on its morphology. The classification is carried out regarding possible QRS patterns and their wavelet transform. We evaluate the algorithm on the MIT-BIH Arrhythmia database. The QRS detector has an average sensitivity of Se = 99.91% and a positive predictivity P+= 99.72% over the first lead of the database.