Analysis of electrocardiograms for subcutaneous monitors

We are developing a subcutaneous cardiac arrest monitor and alarm with electrodes that have spacing of a few centimeters. We hypothesize that closely spaced bipolar electrodes that provide QRS amplitudes of a millivolt or more in sinus rhythm (SR) will not provide similar amplitudes during ventricular tachyarrhythmia (VT/VF), and that an orthogonal set of electrode pairs in diagonal is necessary to produce signals of sufficient amplitude in order to distinguish these rhythms. Methods: Forty patients were studied during ICD implantation in the clinical electrophysiology laboratory. A square array of 9 electrodes with 4- to 10-cm spacing between adjacent electrodes was placed on the patient's anterior chest over the left heart. The center electrode in the left most column is located in the 5th intercostal space at the left sternal border. Eight-channel recordings were made in SR and during induced VT/VF from the 8 peripheral electrodes referenced to the central member of the array. From these recordings all 20 bipolar ECGs of adjacent (including diagonally adjacent) electrodes were constructed algebraically. QRS peak-peak amplitudes in SR and VT/VF were measured in each bipolar lead and in the spatial vector formed by summing the squares of each adjacent and orthogonal pair of these leads. Results: Electrode pairs that yielded optimal ECG amplitudes in SR were not always the pairs that yielded optimal amplitudes in VT/VF. But in every patient two orthogonal pairs could be found whose QRS vector amplitude is sufficient in both SR and VT/VF to separate the rhythms. Conclusions: A patient-defined set of bipolar electrode pairs is suitable for automatic separation of SR and VT/VF by rate in subcutaneous ambulatory monitoring.