Atrial flutter (AFL) circuit localization is a significant information that may affect the difficulty of the catheter ablation procedure. It is usually determined during the procedure itself. Knowledge of this information beforehand could aid clinicians assess and plan the operation in advance to improve efficacy. It is known in the literature that AFL variability can be used to discriminate localization non-invasively. However, respiratory motion may introduce variability into the observation. Due to the position of the right and left atrium in the chest, it may affect left and right AFL differently. This may explain the difference in left and right AFL variability. To address this hypothesis, we analyzed the effect of removing respiratory influence from f wave observations on classification accuracy.
25 records of 12-lead ECG data from AFL patients were used. QRS complexes and f waves were segmented from the signal and transformed into vectorcardiograms (VCG). Respiratory motion, represented as loop scaling and rotation was estimated at the time instant of every QRS using a least squares loop alignment formulation. The motion parameters were interpolated between two consecutive QRS instants before being used to estimate the motion at every f wave time instant. The motion parameters were used to correct each f wave observation. VCG loop parameters were then estimated, and several statistical measures on the set of parameter values were calculated and fed into a classifier.
Bland-Altman analysis was performed to determine the effect of removing respiratory motion on the variability of the VCG loop parameters. The results show that f waves were minimally affected by respiratory motion, and the variability observed in the ECG was intrinsic to AFL. By exploiting this variability, good discrimination between right and left atrial flutter can be achieved (max accuracy of 0.96, corresponding to a sensitivity and specificity of 92% and 100%).