INTRODUCTION

Curative therapy of supraventricular arrhythmias, using radiofrequency catheter ablation, is a minimally invasive procedure with a high success rate and a low complication rate [1]. However, the treatment is time consuming and exhausting for the patient and the medical staff. Additionally, the success rate decreases significantly for complex tachyarrhythmias such as atrial fibrillation (AF) or reentry tachycardias with multiple pathways, whereas complication rate and duration of the intervention increase in such cases.

Within the last years catheter based electro-anatomical mapping (CARTO^TM , Biosense Webster Inc.) has significantly improved the curative treatment of tachyarrhythmias such as atrial flutter and focal atrial tachycardia. In an electro-anatomical map color coded activation times are superimposed on an anatomical model of the target chamber. Thus, electro-anatomical mapping provides an image of the activation sequence in an anatomically correct geometry, which was a major innovation in cardiac electrophysiology.

As a severe limitation of electro-anatomical mapping, each point has to be recorded manually from successive beats. Thus, the approach is time consuming and restricted to stable, regular rhythms. A recent study in humans showed that a similar information as provided by electro-anatomical mapping can be obtained by combining anatomical data assessed by magnetic resonance imaging (MRI) and functional data assessed by electrocardiographic mapping [2]. The activation pattern was computed for an individual model of the ventricles from single beat data. The present study explores the application of this approach to the human atria.

METHODS

Theoretical background

In this section only a short introduction into the method is given. Details can be found elsewhere [2] and [3]. The solution of the inverse problem is based on a multiple solution of the forward problem.

In the forward problem, the electric potential field is computed, applying the boundary element method (BEM). The bioelectric properties of the cardiac muscle are considered applying the bidomain model. Here, the transmembrane potential enters the source term. A priori knowledge on the cardiac action potential time course can be used for obtaining a source pattern from a given activation sequence.

For the inverse problem, an initial guess for the activation sequence is obtained applying the critical point theorem [4]. The activation sequence is modified iteratively that the associated forward solution fits the measured data. Due to the ill-posed property of this inverse problem regularization techniques have to be applied.

Data acquisition

Two patients underwent radiofrequency ablation therapy of a supraventricular tachycardia (patient A, male, 65 Y, atrial flutter and patient B, female, 36 Y, WPW-syndrome, right sided accessory pathway). The study was approved by the local ethics committee and written informed consent was obtained from both patients.

Before the treatment in the catheter laboratory individual anatomical data were obtained, using MRI. The atrial geometry was recorded in CINE-mode during expiration. The lungs and the torso shape were recorded in T1-FLASH-mode during expiration. 11 markers (vitamin E capsules) where used to couple all data acquired geometrically to the MRI frame. From this data a boundary element volume conductor model of the end-diastolic geometry was reconstructed for each patient (see Fig. 1).

Fig. 1: Boundary element model of an individual torso geometry (patient A). The lungs and the endocardial surfaces of the atria and the ventricles are depicted in gray. The chest and the epicard are shown as a wire frame. The electrodes on the chest are indicated by spheres; r, l and f define the Wilson terminal.

The patients were moved to the catheter laboratory, and a 62-channel ECG was recorded during the intervention with a sampling rate of 2048 Hz using the Mark-8 system (Biosemi V.O.F., Amsterdam). A Wilson-terminal defined the reference potential [5]. Signals were bandpass filtered with a lower corner frequency of 0.3 Hz and an upper corner frequency of 400 Hz. Radiotransparent carbon electrodes were used in order to allow simultaneous X-ray examination. Electrode and marker locations were digitized by the Fastrak^®system (Polhemus Inc.). The data was coupled to the MRI-frame in time and space.

For validation purposes an electro-anatomical map (CARTO^TM) of the right atrial endocardium was collected during pacing at the coronary sinus (CS) ostium with a 650 ms time interval for stimulation. Measurements were carried out in the waiting period after successful target ablation. After the intervention the marker locations were digitized in the CARTO^TM reference frame by the ablation catheter and this data was used for coupling the intracardial data to the MRI-frame.

RESULTS

All results were computed from single beat data. The activation time (AT) map for CS pacing in patient A is shown in Fig. 2. In the right atrium (RA) the first onset of activation occurs in the region of the coronary sinus ostium at 13 ms. In the left atrium (LA) activation starts at 4 ms in the posterior wall close to the medial portion of the coronary sinus. The mean geometrical error between the modeled RA and the points of the catheter map was 6.4 mm. The localization error of the first right endocardial breakthrough was 6 mm. The global AT pattern was in good qualitative agreement with the electro-anatomical map.

Fig. 2: The activation time map on the endocardium of right and left atrium for the CS pacing protocol for patient A is shown in a gray shading. The atria are shown in a posterior left-lateral view. (IVC ... inferior vena cava, LL ... left lower pulmonary vein (PV), MA ... mitral annulus, RL ... right lower PV, RU ... right upper PV, SVC superior vena cava).

A similar result was obtained for CS pacing in patient B. The mean geometrical error between the model of RA and the electro-anatomical map was 5.3 mm. The localization error for the first endocardial breakthrough in the RA was 12 mm.

DISCUSSION

The feasibility of noninvasive cardiac activation time imaging was subject of a validation study in humans. Validation was carried out in the RA, which is easier accessible than the LA. Comparison with electro-anatomical mapping was performed for pacing in the CS, which ensured a stable position of the pacing catheter during electro-anatomical mapping. Activation time maps were computed off-line from single beat data. The localization of the first onset of activation was reconstructed with a mean geometrical error of 8 mm. This error was in the same order of magnitude as errors introduced by geometrical coupling. Therefore, we conclude that noninvasive activation time imaging could become a powerful tool in the diagnosis and treatment of supraventricular arrhythmias.

Acknowledgments: This study was supported by the Austria Science Fund (FWF) under grant START Y144-INF. The authors would like to thank Dr. M. Schocke and Dr. Ch. Kremser for support.