Shape description as a key to ECG analysis

The authors present an approach to the automatic analysis of electrocardiograms (ECGs) based on the criteria used by a practicing cardiologist in the visual interpretation of ECGs. Practicing cardiologists primarily exploit shape to focus their attention on the ECG features to be studied in detail. Shape is also an important feature in the description, recognition, and classification of patterns. In order to assist the cardiologist in this activity, a method of producing shape descriptions of the tracing is presented. This method is based on attributed conditional rewriting systems and allows the integration of structural methods and traditional signal processing techniques. The architecture of a system for the integration of these tools and the construction of descriptions is presented. The use of such a system to set up and manage archives of tracings is discussed.