ABSTRACT

This work presents the development of an automated modeling system to obtain a piece-wise linear electrical model of bradycardia pacing leads. In vivo studies were performed on a canine model under acute conditions, and a computer algorithm was used to determine and verify the model parameters. This iterative algorithm provided a first order electrical model of a type 4056 atrial pacing lead, where the mean point-to-point discrepancy in the time domain was around two percent. This study showed that a piece-wise linear equivalent circuit for the Impedance seen by a pacemaker can be determined automatically from acutely collected data.

BACKGROUND

A linear electrical circuit model for the impedance seen by a pacemaker is needed for the development of the output circuits used in brady pacemakers. Such a circuit would simulate the lumped impedance created by the pacing lead and the tissue during and after delivery of the pacing pulse. This model would provide a realistic estimate of the electrical load for the designers of the pacemaker output circuitry.

FIG. 1. DATA COLLECTION BLOCK DIAGRAM

DATA COLLECTION

Data were collected using a simple PCbased data acquisition system under acute conditions in an animal laboratory, as seen in Figure 1. Two pacing pulses, both -3.0 volts in magnitude and 1.0 msec in pulse width, separated by 50 milliseconds, were delivered to the canine atrium in @ipolar fashion. The D/A converter and the pacing lead were coupled with a 10 AF coupling capacitor which is fairly typical for modern pacemakers. Unipolar recordings were made between the tip and the indifferent electrodes, starting 5 milliseconds before the first pace and continuing for a total of 100 milliseconds at a rate of 1000 samples per second. The surface ECG verified that the first pace did capture the heart and the second pace was delivered during the atrial refractory period.

DATA ANALYSIS

Data analysis of the recorded signals was performed offline using computer software where the user enters the desired electrical network of the model and initial estimates of the component values. Linear passive components and sources were allowed for the model network. A SPICElike custom procedure simulated the model circuit to determine its output to the pacing stimuli. The am of the squared errors was used to determine the discrepancy between the actual data and the model output. The component values in the model were then updated using a multidimensional downhill simplex method called AMOEBA [1]. This process continued to iterate until the am of the squared errors was below an acceptable level.

RESULTS

Figures 2 and 3 show the equivelent circuit of the experimental setup and the circuit determined by the model during data analysis, respectively. Since the coupling capacitor is a part of the pacer, the actual impedance seen by the pacer consists of R1, R2,C1, and a DC voltage source, hence it is a first order model. Figure 4 shows the time domain traces of the actual measured tip-to-indifferent signal as small squares and the output of the model as a continuous line. As can be seen from this figure, the actual measured signal and model outcome are quite close. Point to point error measurement showed that the mean discrepeancy was 2.0 +- 2.9 percent.

Figure 2. Experimental Setup

Figure 3. Developed Model

Figure 4. Electrode voltage versus time

CONCLUSION

This study provides an automated method to obtain linear models for the electrical loads of pacing circuits using computer algorithms. As demonstrated by the above results, a first order model did provide close approximations to the measured data. However, this linear model is valid only for one of the entire domain of pacing applications, and the model parameters are expected to vary as the lead type and stimulus parameters are changed. The method is not limited to the first order model used above, although the use of second order models did not provide significant reductions in the error.

REFERENCE

1. Numerical Recipes in Pascal, Press, WH, et al, Cambridge University Press, 1992.