Data fusion of electrophysiological and haemodynamic signals for ventricular rhythm tracking

Demonstrates how the cause of the ventricular contraction (VC), reflected by its electrophysiological precursors (ECG), and its related effect on the haemodynamic system [characterized by the systemic arterial pressure (SAP), sometimes called the mean systolic arterial pressure (MSAP)] can be fused together for a ventricular rhythm (VR) tracking purpose. Data fusion of heterogeneous signals, such as ECG and SAP, appears to be an interesting alternative to overcome the relative drawbacks of signal processing on univariate channels. Behind the clinical aim of providing a more robust VR signal, this article exemplifies how to manage such heterogeneous data when asynchronous events represent the same VC. From an engineering perspective, the main objectives of this article are, firstly, to present a formal representation of the problem of VR tracking and, secondly, to introduce a novel signal processing architecture that is able to manage and fuse cardiovascular data that are both asynchronous and heterogeneous. The evaluation of the performance of this signal processing system uses a very noisy data set extracted from the IMPROVE data library