Improvement in Automated STEMI Detection by Modeling and Classification of the ST Segment

Accurate analysis of the ST segment is critical in diagnosis of the ST segment elevation myocardial infarction (STEMI). Negligible to high levels of artifact from different sources may be present throughout the ST segment and could either mimic a nonexistent ST elevation (false STEMI), or change the actual ST level and negatively impact the precise diagnosis of the underlying disease. Filtering and averaging the beats in an ECG interval reduces the artifact level on the ST segment, but does not eliminate it completely. We studied two approaches for improving the ST segment analysis, a short-duration smoothing filter and an ST segment curve-fitting model, and compared them to the raw ECG average beats. The smoothing approach used polynomial least-squares approximation of the average beat ST segment in a 20 msec window. The curve-fitting method modelled the ST segment by a section of a fitted parabola using a quadratic polynomial equation. The 12-lead 10-second records from a test database collected in a single medical center and annotated by the experts were analyzed for STEMI detection. Smoothing or curve-fitting the ST segment reduces the number of false STEMI detections significantly compared to the raw ECG average beats.