Robust beat detection on noisy differential ECG

• Published in: 2016 39th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO) pp. 381 - 386
• Main Authors: Lavric, Primoz, Depolli, Matjaz
• Format: Conference Proceeding
• Language: English
• Published: Croatian Society MIPRO 01.05.2016
• Subjects: Algorithm design and analysis; Complexity theory; Detection algorithms; Electrocardiography; Estimation; Noise measurement; Smoothing methods
• DOI: 10.1109/MIPRO.2016.7522172

The differential ECG measurement is performed by a small gadget attached to a person's chest. In contrast to the standard 12-channel ECG, which is most often short and measured on a resting subject, the small form factor of the gadget allows for several days long measurement of active subjects. The resulting measurements are not only novel in their time and activity coverage, but they also require novel approach to their analysis. First, ECGs are noisier because the electrodes are placed close together and because measurements are made on physically active subjects. Second, the gadget is not always placed on by a trained professional and can be miss-oriented, producing a wide range of ECG orientations. Finally, the ECGs are measured with low sample ratio which helps conserve battery life of the gadget. In this work, we try to develop the first stage of such differential ECG analysis - algorithms for noise estimation and for beat detection. These novel algorithms have to be able to deal with noisy and sparsely sampled ECGs and with various beat shapes. Furthermore, these algorithms should be able to run in real-time applied to the ongoing measurement and on a very computationally weak portable devices, while maintaining power efficiency. Two algorithms for noise estimation and one for beat detection that fulfill those constraints are presented in this work.