Objective: Radiofrequency (RF) cardiac ablation is a commonly used method for treating cardiac arrhythmias in which the information of the dynamic lesion heating is critical to cardiologists but is currently lacking. Electrical impedance tomography (EIT) is a temporal modality of imaging the changes in the electrical properties within a measured object and hence might be able to track the electrical variation due to temperature changes within the myocardium. Within this paper, (1) a time-efficient algorithm with self-weighted NOSER-prior and (2) a measurement filtering process for optimizing the number of measurement were proposed for monitoring the lesion size during the cardiac RF ablation, taking advantage of internal catheter-based electrodes and the prior information of anatomical structure and the catheter location, which are usually available during the ablation course.
Approach: A tank model with a circular myocardium of 12 mm in thickness, 16 external electrodes on the boundary and three internal catheter-based electrodes positioned inside the endocardium were made. The ablations were simulated using Pennes' bioheat transfer equation and the simulated temperature gradients were then transferred to EIT measurements. The algorithm used one reference ablation for its optimization and then was tested with numerous 90 s ablations containing three disturbances: the catheter location mapping, the wide range of varied myocardium conductivity and the blood's cooling convection, and the Gaussian noises with 10-40 µV in standard deviation.
Main results: The results showed that, with the optimized number of 55 measurements, the algorithm still performed well when dealing with all three disturbances plus the random noises up 25 µV. Specifically, the lesion depth and width were measured within 1.6 mm and 3.2 mm in error respectively in at least 80% out of 100 simulated ablations.
Significance: The algorithm has successfully measured the lesion size with good accuracy and tolerances of noise and other system perturbations. More tests in vitro and in vivo are required in the future to confirm the algorithm's feasibility.