Discrimination between ischemic and artifactual ST segment events in Holter recordings

Abstract ST segment changes provide a sensitive marker in the diagnosis of myocardial ischemia in Holter recordings. However, not only do the mechanisms of ischemia result in ST segment deviation, but also heart rate related episodes, body position changes or conduction changes among others, which are considered artifactual events when ischemia is the target. In order to distinguish between them, the very similar signatures of ST modifications has led us to look for other ECG indices such as heart rate-based indices, correlation between the absolute ST segment deviation and heart rate series, the interval between the T apex and the T end , T wave amplitude, the signal-to-noise ratio and changes in the upward/downward slopes of the QRS complex. A discrimination analysis between the three types of events: ischemia, heart rate related episodes and sudden step ST changes (body position changes and conduction changes) has been performed on the Long-Term ST Database, reaching an accuracy of 82.3%. If we focus on distinguishing between different ST signatures, transient episodes (ischemic and heart rate related) and sudden step ST changes, it results in a sensitivity of 76.8% and a specificity of 98.3%. When classifying ischemia from heart rate related episodes, both with a very similar ST level pattern, a sensitivity of 84.5% and a specificity of 86.6% are reached. Finally, for separating ischemia from any other ST event, a sensitivity of 74.2% and a specificity of 93.2% are obtained.