Reconstruction of Limb Leads Recorded on the Subject’S Torso

Judyta Salamon
Warsaw University of Technology


Aims: Standard limbs electrodes positioning may be challenging when a patient is subjected to stress test, due to limbs movement. Mason-Likar electrodes position goes against it, as limbs electrodes are placed on the torso. On the other hand, such recording shows number of differences to standard positioning. We wanted to find a compromise between them, placing limb leads on the human body to avoid motion artifacts and using signal reconstruction to make the record diagnostic.

Methods: 12 lead ECGs were measured by two devices simultaneously. Precordial leads were placed on the standard positions. Limb leads were placed on standard positions (first device, on limbs) and on the side of the chest (second device, in men above and below electrode V5 and symmetrically on the right side, in women right above and below point between electrodes V5 and V6). 79 men and 51 women were recorded. All subjects were healthy, without any cardiological history. Reconstruction has been performed based on conversion matrices, that were created using linear and lasso regression, polynomials fitting and SVM.

Results: For each method of matrix creation, the reconstructed signal has been compared to the one recorded in reference positions. The results for each method have showed results of Symmetric mean absolute percentage error (SMAPE) from around 6% in lead V4 up to 44% in lead III in men and women. Mean difference between all points varies from 0.019 mV to 0.045 mV. For linear regression mean difference between R wave amplitudes was 0.1 mV, 0.03 and 0.08 mV for Q and S waves respectively.

Conclusion: It may be possible in the future to introduce into practice the recording of limb leads from the electrodes on the subject's torso. The main advantage will be the signal diagnostically correct, and at the same time not containing motion artifacts.