NONINVASIVE THREE-DIMENSIONal LOCALIZATION OF ORIGIN
OF CARDIAC ACTIVATION BY MEANS OF A HEART MODEL

G. Li , B. He
Department of Bioengineering, University of Illinois at Chicago, SEO 218, M/C-063,
851 S. Morgan Street, Chicago, IL 60607, USA

Abstract: The performance of a new heart-model-based approach in localizing the site of origin of cardiac activation has been systematically evaluated by computer simulation. Pacing protocols, both single-site pacing in the atria and ventricles, and dual-site pacing in the ventricles, were used to simulate localized myocardial activation at different myocardial areas. The present simulation results show that the average recognition accuracy for localizing the site of origin of cardiac activation over the atria and ventricles was about 3.4 + 1.3 mm under 10 mV potential noise, and about 3.8 ± 1.7 mm under 10 mV potential noise plus 10 mm geometry noise. The present results suggest this new approach may become an alternative in localizing the site of origin of cardiac arrhythmia.

INTRODUCTION

It is of importance in both basic research on the mechanisms of cardiac arrhythmia and clinical applications in guiding radio-frequency catheter ablation, to accurately and reliably determine the site of origin of cardiac arrhythmia. Body surface potential maps (BSPMs) have been used to aid the localization of the site of the accessory atrioventricular pathway before radio frequency ablation or surgery [1]. However, the ability of BSPM for identification of cardiac electrical sources is limited by the smoothing nature of the torso volume conductor. A lot of effort has been made to overcome the smoothing effect by solving the so-called electrocardographic (ECG) inverse problem [2].

Recently, we have developed a new three-dimensional ECG localization technique to solve the ECG inverse problem [3], in terms of the heart-model parameters. Our initial simulation study [3] has demonstrated that the new technique can provide accurate localization of the site of origin of ventricular activation. As a further step to facilitate future experimental study and clinical applications, in the present study, we have extensively evaluated the performance of the new heart-model-based localization approach in resolving pacing sites under different conditions by computer simulations.

METHODS

Basic Principle

In the present study, the ECG inverse solution in terms of heart model parameters is obtained by using a parameter optimization system [3].  After a site of cardiac activation in the heart model is initialized by a preliminary diagnosis system, a multi-objective nonlinear optimization procedure is implemented to estimate site of origin of cardiac activation. Three objective functions of the optimization system were constructed based on 3 characteristics in the BSPMs [3]. The target site of the optimization procedure is localized when the objective functions, that assess the dissimilarity between the measured and simulated BSPMs, are minimized. The heart model parameters corresponding to the site of origin of cardiac activation are considered to be the inverse solution in the present approach. The recognition accuracy is evaluated by the localization error, defined as the Euclidean distance between the “true” pacing site and the inversely estimated pacing site.

Simulation Protocols

A previously developed computer heart-torso model [4] was used in the present simulation study, in which the heart model has approximately 65,000 myocardial cell units with a spatial resolution of 1.5 mm. Three pacing protocols, single-site atrial pacing, single-site ventricular pacing, and dual-site ventricular pacing, were used to evaluate the performance of the present ECG inverse approach. For the cases of single-site pacing, 26 myocardial cell units throughout the atria or ventricles were randomly selected as the “true” pacing sites. For the case of dual-site pacing, 26 pairs of myocardial cell units in a myocardial region adjacent to the atrioventricular ring were randomly selected as the “true” pacing sites. For each paced activation sequence, 200-lead BSPMs were calculated using the boundary element method.

In order to assess the clinical applicability of the present technique in localizing the site of origin of cardiac activation, Gaussian white noise (GWN) with 4 different levels was added to the BSPMs and 4 different lead configurations were used to evaluate the effects of measurement noise and electrode number. Furthermore, the effects of the following three factors on the recognition accuracy were also evaluated: torso geometry uncertainty, heart position uncertainty, and electrode placement uncertainty. To study the effect of torso geometry uncertainty, two modified torso models, obtained by respectively enlarging and reducing the standard torso model by 10% were used to replace the standard torso model in the forward computation of the BSPMs. The heart position uncertainty was simulated by shifting the heart position in the heart-torso model along x-direction (from the right to the left, RSX/LSX) by 10mm. The electrode placement uncertainty was simulated by adding geometrical GWN of 5 mm and 10 mm to the coordinates of the electrodes.

RESULTS

Fig. 1 illustrates three representative examples of the inversely estimated pacing site when 10 mV GWN was added to the simulated BSPMs. Fig. 1(a) shows an example of the single-site pacing in the lateral wall of the right atrium with the localization error of 3.35 mm. Fig. 1(b) shows an example of single-site pacing in the anterior region of the middle of the right ventricle with the localization error of 3.0 mm. Fig. 1(c-d) show an example of dual-site pacing in the lateral region of the right ventricle and in the lateral posterior region of the left ventricle with localization error of 3.35 mm and 2.12 mm, respectively.

 

Figure 1. Three representative examples on comparison between true pacing site (TPS) and inversely estimated pacing site (ISS). (a) an example of the single-site atrial pacing; (b) an example of the single-site ventricular pacing; (c) and (d) an example of the dual-site ventricular pacing.

Figure 2. The effects of potential noise levels and number of electrodes on the localization error (LE). (a) The average effects of potential noise levels; (b) The average effects of number of electrodes.

The effects of potential GWN and number of electrodes on the recognition accuracy of the pacing site estimation are illustrated in Fig. 2. The average localization error among all 3 protocols, under 4 GWN levels is shown in Fig. 2(a). Note that high recognition accuracy is obtained even under 20 mV GWN with the mean localization error less than 4 mm. The average results across all 3 pacing protocols, under 4 lead configurations are shown in Fig. 2(b). Note that when only 32 electrodes are used, a dramatic decrease in recognition accuracy is obvious as compared with other 3 lead configurations.

The averaged localization error (mean±standard deviation) corresponding to 3 pacing protocols and 3 uncertain factors are summarized in Table 1. 10 mV potential GWN was added to the BSPMs in each case. Compared to the standard heart-torso model whose average localization error is 3.4±1.3 mm, only limited increase in localization error is observed for the 3 uncertainties. In another word, the present simulation suggests that the recognition accuracy is relatively insensitive to the heart-torso geometry model and geometry noise.

TABLE I
Effects of 3 uncertainties on recognition accuracy

Torso-Geometry Uncertainty			Heart-Position Uncertainty		Electrode -Placement Uncertainty
ST	ST-	ST+	RSX	LSX	5 mm	10 mm
3.4±1.3	4.1±1.8	4.2±1.7	3.7±1.8	3.7±1.6	3.6±1.5	3.8±1.7

Note: ST: Standard torso, ST-: ST-10%, and ST+: ST+10%.

DISCUSSION

In the present study, we have conducted comprehensive simulation studies to evaluate the performance of the heart-model-based localization approach in localizing site of origin of cardiac activation. The promising results obtained in this study suggest this approach may become an important alternative for localizing the site of origin of cardiac arrhythmia. Further investigations will be conducted in an experimental and clinical setting to directly assess the clinical validity of the present approach in localizing the site of origin of cardiac arrhythmia.

Acknowledgments: The authors are grateful to Jie Lian for useful comments on the manuscript. This work was supported in part by an American Heart Association Established Investigator Grant AHA0140132N and NSF CAREER Award BES9875344.

REFERENCES

[1] M. Dubuc, R. Nadeau, G. Tremblay, et al. “Pace mapping using body surface potential maps to guide catheter ablation of accessory pathways in patients with Wolff-Parkinson-White syndrome,” Circulation. 1993;87:135-143

[2] R. M. Gulrajani. “The forward and inverse problems of electrocardiography,” IEEE Eng. Med. Biol. Mag., vol. 17, pp. 84-101, 1998.

[3] G. Li, and B. He. “Localization of the site of origin of cardiac activation by means of a heart-model-based electrocardiographic imaging approach,” IEEE Trans. Biomed. Eng., vol. 48, pp. 660-669, 2001.

[4] W. Lu, Z. Xu, and Y. Fu. “Microcomputer-based cardiac field simulation model,” Med.  Bio. Eng.  Comput., vol. 31, pp.384-387, 1993.