A novel method for ECG signal morphology analysis using tortuosity estimation

Objective: The present work aims to evaluate an electrocardiogram (ECG) signal with a new algorithm based on the tortuosity of the signal, which seeks a better performance in differentiating pathologies without losing signal information due to abnormalities. Methods: The digitized ECG signal is studied as a set of segments that make up the complete signal. For each segment, the associated slope is calculated and, using the chain code, the change in the slope between contiguous segments is obtained to conform the tortuosity of the signal. This procedure was used for 12-lead ECG signals for 14 subjects from the arrhythmia group and 14 subjects form the control group, resulting in 168 signals analyzed per study group. Results: The tortuosity results between both study groups were compared using the statistical t-test, showing a significant difference (p < 0.05) in 9 of 12 leads of the ECG signal. This was compared with the RR-interval values, in which no significant difference was found in any lead between the study groups, indicating better discrimination performance with the proposed algorithm. Conclusions: The proposed method can differentiate for most of the ECG leads between pathological and healthy signals. Significance: This method was compared to the widely used method RR-interval estimation and outperformed it by obtaining statistical differences in 9 more leads between the study groups.