Computer modelling and interpretation of paced electrocardiograms for analysis of pacemaker function.
Saul E. Greenhut · Deep Blue (University of Michigan) · 1991
Electrocardiographic (ECG) analysis is critical to pacemaker patient follow-up for identifying pacemaker function. Because of the complexity of manual diagnosis of the paced ECG, sophisticated computer analysis is proposed to aid the interpretation of pacemaker malfunction. An automated ECG analysis algorithm for dual-chamber pacemakers has been developed using pacing stimulus onset, atrial and ventricular depolarization onsets, intervals, and morphologies. The interpretation algorithm defines ECG cycles and pacemaker operating mode for each cycle. Appropriate analysis of atrial and ventricular output, capture, and sensing is performed and specific interpretations given. The algorithm is unique in the utilization of intrinsic pacemaker logic, atrial depolarization timing, and waveform morphology in the decision process. Pacemaker blanking, refractory, and other periods are incorporated into the analysis as well as tolerances in device activities and cardiac chamber activation. Software is implemented as a rule-based structure which includes a pacemaker description language that allows a variety of pacemaker model analyses to be incorporated with minimal effort. A new heart-pacemaker interaction (HPI) model was developed to serve as a development tool for the interpretation algorithm. The stochastic network model provides a concise framework for simulation of 25 classes of arrhythmias, 13 pacemaker modes, and a unique feature of simulation of a variety of pacing and sensing failures. Seventy-seven simulated paced ECG passages served as the system training set. The computer algorithm was tested by varying pacemaker parameters to simulate failures in five patients with DDD (dual-chamber pacing and sensing) pacemakers in which surface and esophageal ECGs were recorded. Thirty-three 15 second ECG passages were randomly selected which included single or multiple pacemaker failure simulations of output, capture, and sensing in atrial and ventricular chambers. Two errors in automated analysis were detected which yielded an overall computer algorithm success of 99%. This thesis presents a robust computer algorithm for analysis of pacemaker ECGs for the determination of pacemaker function and malfunction. The system was validated using a novel model of heart-pacemaker interaction and clinical ECGs.