Quality estimation of the electrocardiogram using cross-correlation among leads

• DOI: 10.1186/s12938-015-0053-1
• Publication Date: 2015-06-19T14:50:23Z
• Authors: Eduardo Morgado; Felipe Alonso-Ati...; Ricardo Santiago-...; Óscar Barquero-Pérez; Ikaro Silva; Javier Ramos; Roger Mark
• Keywords: Binary classification

Fast and accurate quality estimation of the electrocardiogram (ECG) signal is a relevant research topic that has attracted considerable interest in scientific community, particularly due to its impact on tele-medicine monitoring systems, where ECG collected by untrained technicians. In recent years, number studies have addressed this topic, showing poor performance discriminating between clinically acceptable unacceptable records.This paper presents novel, simple algorithm estimate 12-lead exploiting structure cross-covariance matrix among different leads. Ideally, signals from leads should be highly correlated since they capture same electrical activation process heart. However, presence noise or artifacts covariance these will affected. Eigenvalues are fed into three supervised binary classifiers.The classifiers were evaluated using PhysioNet/CinC Challenge 2011 data. Our best classifier achieved an accuracy 0.898 test set, while having complexity well below results contestants who participated Challenge, thus making it suitable for implementation current cellular devices.