Session P36.4

Improved 12-Lead ECG Reconstruction from Lead Sub Sets by Dynamic Selection of Frontal Leads

SP Nelwan*, TB van Dam, SH Meij

Erasmus Medical Center
Rotterdam, Netherlands

In recent years, patient monitoring with lead subsets of the 12-lead ECG and subsequent reconstruction of the unrecorded leads has been commercially available. In general, implementations of reduced lead sets use the independent frontal leads I and II and two of the six precordial leads to reconstruct the four unrecorded ECG leads.
However, depending the average direction of the ventricular activation in the frontal plane, QRS amplitudes in leads I or II may be diminished and may have an effect on the signal to noise ratio of the reconstructed leads. It is hypothesized that reconstruction could be improved by selecting a different combination of frontal plane leads. The aim of this study was to develop a frontal lead selection method to improve ECG reconstruction.
Methods: We used a dataset of 2372 diagnostic 12-lead ECGs obtained from subjects with chest pain suggestive of acute myocardial infarction. The ECGs were analyzed by the Modular ECG analysis system and averaged beats were computed. The recordings were divided into a separate learning and test set. General reconstruction coefficients for the frontal leads and precordial leads V2 and V5 were computed from the learning set. Reconstruction performance was evaluated by average Correlation Coefficient (CC) and root mean square error (RMSE).
Results: Of the 1172 patients in the test set, the average heart rate was 82 (SD: 24) bpm, frontal QRS difference was 38 (SD: 51) degrees. A total of 485 (41.3%) ECGs showed a rightward frontal QRS axis (>60 degrees) of which 14 (1.2%) showed a diminished R wave. By applying the dynamic lead selection method, average CC increased from 0.931 to 0.938 and RMSE decreased from 118 to 112 micoVolt.
Conclusion: The dynamic frontal lead selection method increases reconstruction performance and signal to noise ratio in the reconstructed leads.

(Abstract Control Number: 74)