Automated detection of the culprit artery from the ECG in acute myocardial infarction

The aim of this study was to develop, implement and evaluate criteria for automated detection of the culprit artery in patients with an acute myocardial infarction. ECG and PCl data was retrospectively gathered in Zealand, Denmark, from patients who had presented with symptoms suggesting an acute coronary syndrome, had a prehospital ECG recorded, had PCl on the same day as the ECG recording, and were subsequently identified as having single vessel disease. 307 patients were selected (218 male, 89 female, mean age 61.8 ± 12.3 years) as a training set. ECG criteria were designed to locate the culprit artery based on the location of ST deviation. The training set was used to optimise the criteria which were then incorporated into the Glasgow ECG analysis program. The ECGs were analysed using the enhanced software and the suspected culprit artery for each ECG was identified. The sensitivity and specificity of identifYing each type of occlusion was calculated, using the location determined from the coronary angiogram as the gold standard. The SE and SP for a report of LAD was 69% and 94%, for RCA was 64% and 94% and for LCX was 57% and 96%. In conclusion the detection of the culprit artery from an ECG can be automated with an acceptable degree of accuracy.