Electrocardiogram beat type dictionary based compressed sensing for telecardiology application

Effective compression of Electrocardiogram (ECG) is a vital task in telecardiology application. Compressed sensing (CS) offers a low energy implementation based solution to the telecardiology system. In this work, an efficient beat type dictionary based ECG-CS approach is proposed. The main objective of this study is to incorporate the advantages of both beat type dictionary and non-uniform random sensing matrix for effective patient-agnostic based signal recovery. Unlike patient-specific dictionary based CS approaches, the proposed beat type dictionary offers high-quality signal recovery without the training stage for individual ECG record. The performance of the proposed scheme is evaluated using the standard MIT-BIH database. The quantitative performance matrices such as compression ratio (CR), percentage root mean square difference (PRD1), root mean square error (RMSE), signal to noise ratio (SNR) are compared with the existing CS approaches to quantify the efficacy of the proposed scheme. At PRD1 of 9%, the proposed beat type dictionary-based method presents 33.5% more CR than adaptive dictionary-based CS approach. An in-depth analysis of the results highlights that the proposed beat type dictionary based CS scheme offers an efficient solution to the patient-agnostic based signal recovery and can be served as a potential component in the computer-based automated medical system.