Complex-Pan-Tompkins-Wavelets: Cross-channel ECG beat detection and delineation

Abstract The Electro Cardiogram (ECG) provides insight into the different phases of a heart beat and various kinds of disorders which may affect them. For the identification and treatment of these conditions it is crucial to properly detect each heartbeat and delineate P-, QRS and T-waves. The presented Complex-Pan-Tompkins-Wavelets (CPTW) algorithm aims at detecting and delineating heart beats in real-time across any number of channels between one and 64 sampled between 256 Hz and 4.8 kHz. It merges three well established single channel algorithms, the complex-lead by Christov, the Pan-Tompkins and the discrete dyadic wavelet-transform, such that the shortcomings of one algorithm are compensated by the strength of the other. A first study testing the CPTW algorithm was conducted using 75 records of 30 min duration provided by the INCART database. An initial implementation in Python 3 allows to localize and detect QRS complexes with an average sensitivity of 99.57% and a precision of 99.58% could be achieved. The average time required to process a single data set thereby was 12 min. In a second test which included 3 recordings of 3 min duration the scalability of the algorithm with respect to number of channels and sampling rates was accessed. Incrementing the number of channels by a factor of 5.2–62 channels resulted in an 3.1 fold increment in run-time. Raising the sampling rate from 256 Hz to 4.8 kHz elongated the run-time by a factor of just 3.2.