Challenging QR research issues in Electrocardiology

Electrocardiology is a stimulating and promising application domain for Qualitative and Model-based Reasoning research. Much of the research effort has been so far aimed at the automated interpretation of electrocardiograms (ECG’s) [4; 14; 11; 13], for which the interpretative rationale is well established. However, a major drawback of ECG’s, which record electrical signals from nine sites only on the body surface, is their poor spatial covering. Some arrythmias, but also conduction anomalies caused by infarcts and ischemia, can be missed by ECG’s. Thanks to the latest advances in technology, a far more informative technique, namely body surface mapping (BSM), is becoming more widely available. In BSM, the electrical potential is measured over the entire chest surface, and more spatial information about the heart electrical activity is available, though the ability to relate visual features to the underlying complex phenomena still belongs to very few experts [12]. Besides body surface maps, also epicardial or endocardial maps are becoming available, either measured by means of endocavitary probes or obtained non invasively from body surface data through mathematical inverse procedures. These kinds of maps are the most precise in locating the anomalous conduction sites [5]. While a rationale for the interpretation of electrocardiographic maps is progressively being defined, the goal of bringing such techniques to the clinical practice gets closer. In this context, an important role can be played by QR methodologies for spatial/temporal reasoning, that could 1) support the expert electrocardiologist in identifying salient features in the maps and help him in the definition of an interpretative rationale, and 2) achieve the longterm goal of completely automating map interpretation to be used in a clinical context. The delivery of such a tool would be of great impact on health care as cardiac map interpretation is essential for the localization of the ectopic sites associated with ventricular arrythmia. At the current status of research, map interpretation mainly deals with activation isochrone maps and sequences of isopotential maps. The former ones deliver a lot of information about the wavefront structure and propagation: each contour line aggregates all and none but the points that share the same excitation state, and subsequent increasing isocurves are nothing but subsequent snapshots of the travelling wavefront, as it is illustrated by Figure 1; whereas the latter ones give temporal information about electrical phenomena, such as current inflows and outflows.