Detection of microbubble formation during radiofrequency ablation using phonocardiography

Abstract

Aims To detect and characterize the acoustic energy generated by microbubble (MB) formation in an isolated tissue preparation. MB formation during radiofrequency (RF) ablation indicates excessive tissue heating and may precede explosive ‘pops’. Currently, MB formation can only be detected with echocardiography. We hypothesized that MB formation can be detected with high-sensitivity phonocardiography.

Methods and results In a saline bath, RF lesions were created in sections of porcine left ventricle, using a 4 mm tip irrigated catheter. MB formation was visualized with an echocardiography probe. In 20 preparations, RF energy was begun at 25 W and increased by 5 W every 20 s until a pop occurred. A high-sensitivity computerized phonocardiography transducer with frequency bandwidth of 2 kHz and system noise −90 dB (SonoMedica, Inc., Vienna, VA, USA) was coupled to the external glass wall of the bath. In 15 of 20 (75%) preparations, a characteristic acoustic signature corresponding to MB formation was noted before the pop. These signals were within the 600–2000 Hz range and had an intensity range of 10–40 dB. The earliest MB and acoustic signals occurred 51.3±51.5 s before the pop. The acoustic signals continued intermittently up to 10.3±12.9 s before the pop.

Conclusion The acoustic energy created by MB formation can be detected in an isolated tissue preparation, using a computer-based phonocardiography system. Characteristic acoustic signatures are present before pops and correspond to MB formation. Acoustic monitoring for MB formation may allow for the titration of cooled RF ablation without echocardiography.

Introduction

Microbubble (MB) formation and explosive steam ‘pops’ during radiofrequency (RF) ablation are representatives of excessive tissue heating. During RF ablation procedures for atrial fibrillation, tissue overheating may result in pulmonary vein stenosis or fatal atrialoesophageal fistula formation.^1,2 By monitoring for MB formation in RF ablation procedures, the likelihood of complications can be reduced.¹ Currently, MB formation during clinical procedures can only be detected with intracardiac echocardiography (ICE). The use of ICE adds to the expense and complexity of the procedure, and also, ICE is not universally available in all laboratories. Hence, we sought to identify a simpler method to detect MB formation during ablation. The purpose of this study is to test the hypothesis that MB formation can be detected with high-sensitivity phonocardiography because of the acoustic energy produced by the phase change from liquid to gas.

Methods

Experimental preparation

A schematic of the experimental preparation is shown in Figure 1. Porcine hearts were obtained from a local meat-processing facility. Rectangular pieces (2×2×2 cm³) of left ventricle myocardium were pinned to the floor of a tissue bath filled with room temperature-normal saline solution. A 4 mm tip irrigated ablation catheter (Chili catheter and model 8004 RF generator, Boston Scientific, Natick, MA, USA) was positioned perpendicular to the endocardial side of the tissue. The ablation electrode was internally irrigated with saline at 36 mL/min. The RF current was delivered in a unipolar power-controlled fashion from the catheter tip to a ground in the bath. The impedance of the saline solution measured 70±5 Ω between the ablating electrode and the grounding wire. A 3.5 MHz ultrasound probe (Hewlett Packard Sonos 2500, Palo Alto, CA, USA) was immersed in the bath continuously to image bubble formation at the interface between the catheter tip and the tissue sample. The phonocardiography circular sensor, with a diameter of 3.5 cm was coupled against the exterior wall of the bath. This phonocardiography system uses high-sensitivity piezoelectric transducers that input into a computer-based acquisition and analysis module. The processing settings for the frequency range and resolution were set according to the Nyquist theorem. The sampling rate was 44 100 Hz, with a decimation ratio of 11, and a fast Fourier transform size of 1024 (samples). In the post-processing mode, the resulting time resolution was 23.22 ms. The dynamic range of the system was 90 dB. A second sensor was placed close to the apparatus and served as the ambient sensor to record external noise. Input from the external sensor was subtracted from the data sensor to remove ambient noise from the final data.

In each trial, the RF energy was begun at 25 W and was manually increased by 5 W every 20 s until an audible pop occurred. Energy delivery was terminated after the audible pop occurred. The echocardiography image was monitored continuously during the experiment. In the preparations, the occurrence of MBs was related to tissue temperature only, not electrode temperature, power settings, or impedance.³ When MB formation was visually detected, the researcher recorded the corresponding time on the phonocardiography data-acquisition software. Two patterns of MB formation were identified, as previously described.¹ Type 1 MB formation is characterized by intermittent and scattered bubble formation. Type 2 MB formation is characterized by continuous and vigorous bubble formation. The phonocardiography signals were then assessed offline.

Data analysis

For each trial, the phonocardiography spectrograph was analysed for sound patterns including frequency range and intensity. The spectrograph displays time on the horizontal axis and frequency on the vertical axis, and sound intensity at each frequency is indicated by a colour code. The Blackman window was used because it is suitable for audio signals with high-frequency components. The spectrograph was analysed at the times corresponding to the characteristic line patterns of MB formation for each experiment. Intensity (dB) and frequency (Hz) characteristics of these lines were noted. The intervals between MB formations were also assessed for noise or ambient signals that may simulate MB formation. The graphic display and analysis used custom software (SonoMedica Transmedica).

Results

Twenty experiments producing pops were analysed. Of this group, 15 preparations (75%) had a distinct acoustic signal that corresponded to the time of MB formation by echocardiography (Figure 2). In five preparations, MB formation was visualized on echocardiography with no clear acoustic pattern noted. The first MBs and acoustic signals occurred 51.3±51.5 s before the pop. This discrete acoustic pattern recurred intermittently, with the last acoustic signature occurring 10.3±12.9 s before the pop. The last acoustic signature most often preceded the pop by 4 s or less. Overall, the MB acoustic signals were within the frequency range of 600–2000 Hz and had an intensity range of 45–85 dB. This frequency range included the fundamental frequency and the harmonics imposed by the salt bath. Frequency range for the steam pops spanned the whole spectrum from 0 to 2000 Hz because of amplifier saturation and ringing within the glass bath. It was the impression of the investigators that Type 1 bubble formation was associated with higher acoustic frequencies than Type 2. Inability to distinguish accurately the fundamental frequency from the harmonics severely limits the ability to prove this conclusion.

Discussion

The major finding of this study is that MB formation during RF ablation can be detected in an in vitro preparation, using a high-sensitivity phonocardiography system. During cooled RF ablation energy titration is problematic because of the loss of correlation between the ablation electrode temperature and tissue temperatures. MB formation is an indicator of excessive tissue heating, but currently can only be detected by ICE in the clinical setting.^1,3 Potentially, MB formation may be detectable in vivo, using acoustic sensors placed on the patient's chest or incorporated into the tip of the ablation catheter itself.

When a bubble is formed, energy is trapped in it. After formation, the bubble emits a sinusoidal sound that decays as energy is dissipated from the bubble. The acoustic mechanism responsible for bubble sounds is volume pulsation.⁴ The bubble can be thought of as a small compressible area within a region of incompressible fluid, allowing it to oscillate. Relating this to the spring-mass model, the water is the effective mass and the bubble is the spring.⁴ The oscillating bubble in a fluid causes vibrations of low amplitude and these acoustic emissions fall within a frequency range, which is inversely proportional to the radius of the bubble.⁵

In these experiments, an acoustic signature was evident in most but not all preparations producing MBs. In addition, although the MB formation was at times continuous, the acoustic pattern of the MB formation was discrete. These discrepancies may be due to the very high but limited sensitivity of the phonocardiographic system. The acoustic patterns were usually noted to coincide with the onset of MB formation or the transition to a more vigorous MB pattern. It is possible that at these transitions, the release of stored energy in the system is at a maximum and so crosses the sensitivity threshold of the recording system. After this crescendo, the energy associated with ongoing MB formation is below the threshold for detection. For those experiments not producing any acoustic signatures, the energy released by the MBs, even at transitions, may have been below threshold. The wide range of frequencies and intensities that comprise the acoustic signature likely result from a diversity in the size and energy content, respectively, of the MBs being formed at any instant.

The apparent differences in the frequency ranges for Type 1 and Type 2 MB formations are speculative. The acoustic frequency accompanying bubble formation is inversely related to the radius of the bubble.⁴ If the MBs associated with Type 2 MB formation are larger, a lower frequency results. A larger bubble size could result from higher temperature or from the energy content of the liberated gases.³

Limitations

The experiments were conducted in vitro as a test of the principle that acoustic recordings may be used to detect MB formation. This apparatus imposes resonance frequencies in the sound recording, which restrict the analysis. The ability to record acoustic sounds of MB formation in vivo is not known and may be complicated by respiratory, cardiac, and muscle acoustic artefacts.

References