Effects of Acoustic Insonation Parameters on Ultrasound Contrast Agent Destruction

Ultrasound in Med. & Biol., Vol. 34, No. 8, pp. 1281–1291, 2008 Copyright © 2008 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/08/$–see front matter

doi:10.1016/j.ultrasmedbio.2007.12.020

● Original Contribution EFFECTS OF ACOUSTIC INSONATION PARAMETERS ON ULTRASOUND CONTRAST AGENT DESTRUCTION CHIH-KUANG YEH and SHIN-YUAN SU Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu, Taiwan (Received 8 April 2007, revised 15 December 2007, in final form 21 December 2007)

Abstract—Ultrasound contrast agents (UCAs) are used to enhance the acoustic backscattered intensity of blood and thereby assist the assessment of blood perfusion. Characterization of UCA destruction provides important information for the design of contrast-assisted perfusion imaging. High-speed optical observation of single microbubble destruction during acoustic insonation has been performed in previous studies. The results identified that pressure, center frequency and transmission phase have significant effects on the fragmentation threshold. We proposed an acoustic-based experiment method to demonstrate the relationship between different acoustic exposure conditions and the degree of UCA destruction. The method also provides a simple and convenient way to determine the microbubble destruction threshold. The experiments introduced three insonation parameters, including acoustic pressure (0 to 1 MPa), pulse frequency (1, 2.25, 5 and 7.5 MHz) and pulse length (1 to 10 cycles). The term of surviving percentage (SP) was proposed to represent the ratio of UCA backscattered power with and without acoustic insonation. The results showed that the SP decreased with decreasing pulse frequency, but with increasing transmission acoustic pressure and pulse length. In addition, there was an exponential relationship between SP and acoustic pressure, and thus the UCA destruction pressure threshold could be predicted from the fitted exponential curve. The results also show that the degree of UCA destruction was not related to mechanical index (MI). Potential applications of this method include UCA high-resolution destruction/replenishment imaging model, microbubble cavitation, sonoporation in drug delivery and gene therapy. (E-mail: [email protected]) © 2008 World Federation for Ultrasound in Medicine & Biology. Key Words: Ultrasound contrast agents, Surviving percentage, Mechanical index.

mitted pulse. The nonlinear UCA imaging modalities developed to date include phase-inversion, subharmonic and ultraharmonic methods (Lotsberg et al. 1996; Shi et al. 1999; Simpson et al. 1999). Another important characteristic of UCAs is that they can be fragmented by suitable acoustic excitation. Expanding the microbubble radius on the order of 300% will cause the microbubble to collapse rapidly and separate into small fragments (Bouakaz et al. 2005; Chen et al. 2002; Chomas et al. 2001a). It is well known that UCAs can be destroyed reliably using ultrasound pulses with center frequencies ⬍5 MHz, and Chomas et al. (2001b) found that decreasing the center frequency reduces the threshold pressure required for microbubble destruction. Some studies have applied the fragmentation property of UCAs to estimate blood perfusion. Wei et al. (1998) proposed a well-known blood perfusion evaluation model, the microbubble destruction/replenishment technique, in which contrast agents are destroyed by

INTRODUCTION Ultrasound contrast agents (UCAs) are used to enhance echoes backscattered from blood and thereby assist measurements of blood flow. UCAs are typically shell-encapsulated microbubbles whose solubility in the blood is sufficiently low to ensure that they reach the left ventricle of the heart without significant losses. One of the most important characteristics of UCAs is their nonlinear oscillation, which has been investigated in many experimental and theoretical studies (Chang et al. 1995; Forsberg et al. 2000; Goldberg et al. 2001; Lotsberg et al. 1996; Shi et al. 1999, 2000). Nonlinear oscillations allow tissue and UCA echoes to be distinguished based on their center frequency or response to the phase of the trans-

Address correspondence to: Chih-Kuang Yeh, Ph.D., Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, 101, Section 2, Kuang-Fu Road, Hsinchu, Taiwan 30013, ROC. E-mail: [email protected] 1281

1282

Ultrasound in Medicine and Biology

excitation with lower-frequency pulses, and the subsequent flow of intact UCAs into the sample volume is monitored over time to produce a local estimate of the blood flow velocity. This model has been used to study UCA replenishment in rat tumors (Chomas et al. 2001c) and other disease models (Wei et al. 2001; Yeh et al. 2004). The UCA destruction process can also be exploited to improve the contrast between blood and tissue in images. Frinking et al. (2001) suggested applying a release burst to rupture the microbubbles, and then distinguishing the microbubbles from the background tissue by measuring the decorrelation between the echo signals before and after the excitation. Furthermore, the destruction process provides potential application in acoustic cavitation as well as new techniques for drug delivery and gene therapy (Bloch et al. 2004a). Theoretical and experimental methods have been developed for investigating UCA fragmentation (Bloch et al. 2004b; Chomas et al. 2001a; Porter et al. 2006; Smith et al. 2007). Characterizing microbubble destruction provides important information for the design of contrast-assisted perfusion imaging systems. The dynamic motion of microbubbles during acoustic insonation has been observed optically (Bouakaz et al. 2005; Chomas et al. 2000, 2001b). Images captured by a high-speed camera system have shown the destruction of single microbubbles during the ultrasound compression phase. The experimental results suggested that the resting microbubble radius, acoustic excitation pressure, pulse center frequency and pulse length determine the probability of fragmentation. Increasing the pressure and the pulse length, and decreasing the resting radius and pulse frequency are correlated with an increased probability of UCA fragmentation. Commercial UCAs are generally delivered through a patient IV and circulated through the body. Because a large number of microbubbles pass simultaneously through the imaging sample volume, the threshold for microbubble destruction will be affected by variability in bubble scattering and shadowing effects (Yeh et al. 2003). Thus, an index is needed that accurately predicts the degree of microbubble destruction induced by exposure to a diagnostic ultrasound field. In this study, we used an acoustic-based experimental method and an exponential model to predict the degree of microbubble destruction. The method provides a simple way to determine the microbubble destruction threshold in common laboratory experiments. Different acoustic exposure conditions were considered under clinically realistic diagnostic ultrasound fields, including pulse center frequencies from 1–7.5 MHz, acoustic excitation pressures up to 1 MPa and pulse lengths from 1–10 cycles.

Volume 34, Number 8, 2008

Most clinical investigators consider the value of the mechanical index (MI) to be an indicator of exposure conditions related to UCA destruction (Forsberg et al. 2005; Hirokawa et al. 2002). The MI is proportional to the peak negative pressure of the excitation pulse, and increasing the pulse pressure increases the probability of microbubble fragmentation. However, minimizing the excitation pulse pressure required to induce fragmentation is desirable in clinical applications. Forsberg et al. (2005) evaluated the MI as a predictor of exposure conditions related to the destruction of UCAs in B-mode images obtained using a clinical scanner. They concluded that the MI index does not predict the degree of microbubble destruction and hence should not be used by itself to define the exposure conditions for destroying microbubble contrast agents. Because the MI relates to a nonthermal mechanism that may produce bioeffects (O’Brien 2007), we also investigated the relationship between the MI and the degree of microbubble destruction in this study. This paper is organized as follows. We first describe the setup for the acoustic-based experiments used to estimate the degree of microbubble destruction. Then, the microbubble destruction results under different acoustic exposure conditions are presented. An exponential mathematical model is introduced to predict the relationship between the degree of microbubble destruction and the acoustic pressure. The relationship between the MI and the degree of microbubble destruction is discussed and, finally, conclusions are drawn. MATERIALS AND METHODS Experimental setup An acoustical experimental setup was designed to determine the microbubble destruction threshold under different insonation conditions. Three acoustic insonation parameters were used: (i) pulse pressures varying from 0 –1 MPa; (ii) pulse lengths of 1, 3, 5 and 10 cycles; and (iii) pulse center frequencies of 1, 2.25, 5 and 7.5 MHz. Figure 1 shows a block diagram of the experimental system. The setup consists of two types of single-element transducers: (i) a 25-MHz spherically-focused transducer (model V324, Panametrics, Waltham, MA, USA) responsible for imaging and (ii) four lowerfrequency spherically-focused transducers with center frequencies of 1, 2.25, 5 and 7.5 MHz used for microbubble destruction. The specifications of the transducers are summarized in Table 1. The 25-MHz transducer was fixed at 45 degrees relative to and near the outlet of a 200-␮m inner diameter cellulose tube (Spectrum Labs, Laguna Hills, CA, USA), with the focal region of the transducer aligned onto the tube. The wall thickness of the tube is 16 ␮m and the

Microbubbles destruction ● C.-K. YEH and S.-Y. SU

1283

Fig. 1. Experimental arrangement for the threshold of bubble destruction measurements.

attenuation effect in either imaging transducers or destruction transducers is negligible. The four low-frequency destruction transducers were fixed approximately 5 cm upstream from the 25-MHz transducer on the inlet side of the tube, perpendicular to the tube, and the focus of the destruction transducer was also located inside the tube. Acoustically absorbent rubber was placed at the bottom of the water tank to minimize reflections. A digital-to-analog board (model TE5300, Tabor Electronics, Tel Hanan, Israel) was used to generate 1-, 2.25-, 5- and 7.5-MHz tone-burst pulses comprising 1, 3, 5 and 10 cycles, and these drove the transducers via a radiofrequency (RF) power amplifier (model 150A100B, AR, Souderton, PA, USA). The peak negative pressures generated by the four destruction transducers were calibrated to an accuracy of 0.1 MPa using a wideband hydrophone (model HNP-0400, ONDA, Sunnyvale, CA, USA) before performing the experiments. Note that the

Table 1. Transducer characteristics Model

Frequency

Element Size

Focal Length

–6 dB Bandwidth (%)

V303 V304 V308 V321 V324*

1 2.25 5 7.5 25

12.7 25.4 19.1 19.1 6.4

18.5 71.3 72.1 35.6 12.7

78.2 74.9 59.4 94.1 50.6

Unless otherwise noted, the unit of measure is in millimeters. * Imaging transducer.

maximum transmitted pressure of the 1-MHz transducer was 0.7 MPa, and the 7.5-MHz transducer was 0.8 MPa. The destruction pulses were transmitted with a pulse repetition frequency (PRF) of 50 Hz, and the primary and secondary radiation forces were insignificant and hence could be ignored (Dayton et al. 1997). The radiation force is related to the flow and bubble concentration in the acoustic filed. The microbubbles suffer less influence of radiation force under lower PRF insonation pulses. In each combination of frequency, pulse duration and acoustic pressure amplitude, there were 10 independent repetitions experiments and the total exposure duration was 6 s. After a delay of 5 s after excitation with the lowerfrequency pulses, the 25-MHz transducer was operated in a pulse/echo mode to detect the presence of bubbles in the tube. The signals were received by a pulser/receiver (model 5900PR, Panametrics) operating in the pulse/ echo mode via a diode limiter/transformer diplexer circuit at a PRF of 500 Hz. The peak-negative pressure level of the 25-MHz transducer was 0.24 MPa at the focus. The passband of the pulser/receiver hardware filter was from 10 –50 MHz. The energy of the excitation imaging pulse was 4 ␮J, and the receiver gain was 26 dB. A preamplifier connected upstream of the pulser/receiver (model AU-1114-BNC, Miteq, Hauppauge, NY, USA) amplified the RF signal by 31 dB, which was then digitized by a 120-MHz, 14-bit analog-to-digital board (model PCI-9820, ADLINK Technology, Taipei, Taiwan) and stored in M-mode (i.e., motion mode) format

1284

Ultrasound in Medicine and Biology

on a personal computer (Hewlett-Packard, Palo Alto, CA, USA). MATLAB (The MathWorks, Natick, MA, USA) was used to analyze the data off-line. M-mode format means that each vertical line corresponds to the received signal along the axial direction. Such signals are obtained repetitively at a rate defined by the PRF and are stacked along the horizontal axis. A commercial contrast agent (Definity, Bristol-Myers Squibb Medical Imaging, N. Billerica, MA, USA) was used at a concentration of 6 ␮L/10 mL water (3 mL/5 L blood volume). Definity microbubbles have diameters ranging from 1.1–3.3 ␮m and a maximum particle size of about 20 ␮m, with over 98% of them ⬍10 ␮m. Note that each microliter of the Definity suspension contains the 1.2 ⫻ 1010 and 4,800 microspheres before and after dilution, respectively. The dose of Definity usually given to human patients is 10 ␮L/kg by IV bolus injection. A syringe pump maintained the flow velocity of the UCAs solution through the tube at 8.9 mm/s, corresponding to a flow rate of 1 mL/h. Because the lateral beamwidth of the 25-MHz transducer is approximately 120 ␮m, there were fewer than 15 microbubbles within the sample volume. The backscattered signal power received by the 25-MHz transducer was proportional to the number of microbubbles within the sample volume at the pressures (i.e., 0.24 MPa at focus), pulse lengths (i.e., 4 to 5 cycles) and frequencies

Volume 34, Number 8, 2008

Fig. 2. (a) RF backscattered signal. (b) RF M-mode image. (c) Postwall-filtered envelop-detected M-mode image. (d) Sum of the postwall-filtered power over the time interval from 0–0.6 s in (c).

(i.e., – 6dB bandwidth of 19 to 32 MHz) used in this study. In other words, nonlinear excitation is negligible for the 25-MHz imaging transducer. To consider the condition of first-pass concentration of microbubbles in

Fig. 3. Postwall-filtered power as a function of acoustic pressure for a 1-MHz transducer with (a) 1-, (b) 3-, (c) 5- and (d) 10-cycle transmitted pulses.

Microbubbles destruction ● C.-K. YEH and S.-Y. SU

1285

Fig. 4. Postwall-filtered power as a function of acoustic pressure for a 2.25-MHz transducer with (a) 1-, (b) 3-, (c) 5and (d) 10-cycle transmitted pulses.

vivo being higher, the concentrations used in these destruction threshold experiments were higher than those normally used clinically. The destruction threshold depended on the concentration of the microbubbles, and microbubbles at high concentrations tend to be much less susceptible to destruction by ultrasound (Klibanov et al. 1998). Signal processing One M-mode image comprising 512 samples (i.e., 3.2 mm) was acquired using 3,000 firing pulses (i.e., total 6 s) for each exposure setting and divided into 10 groups of 300 pulses for statistical analysis. The acquired data were first filtered with a second-order high-pass Butterworth filter with a cutoff frequency of 20 Hz over a slow time index to remove the stationary echoes from the wall of the tube, as well as wall reverberation echoes that appeared directly below the tube. Figure 2a shows the acquired RF signal, where the two dashed-line boxes indicate the signals backscattered from the front and rear walls of the tube. Figure 2b and c show the RF and postwall-filtered envelop-detected M-mode images, respectively. Note that envelop-detected image means to filter out the carrier frequency component and to keep only the magnitude of the remaining signal. The postwall-filtered power, defined as the square

of the envelop-detected signal amplitude, was used to assess the microbubble destruction threshold in our experiments. The microbubbles were considered to be completely destroyed when the echo power received from the imaging pulses was ⬍5% of that recorded before microbubble destruction. The ideal received power would be zero when the microbubbles were destroyed completely, but there was always a low residual signal because of background noise. Because microbubble echoes are detected far above their linear resonance frequencies, and the microbubbles are small compared with the wavelength at 25 MHz, we assume that they behaved as Rayleigh scatterers. Thus, the integrated backscattered signal power was proportional to the number of intact microbubbles within the tube. The postwall-filtered power was first integrated over a time interval from 0 – 0.6 s, as shown in Fig. 2d, and the resulting curve was integrated over depth. The integration was performed over a region windowed in depth to reject duplicate echoes because of reverberation. For consistency, the same windowed region was used for all process. To describe the degree of microbubble destruction, we also considered the surviving percentage (SP) index, which was defined as the ratio of integrated backscattered power estimates with and without destruction pulse insonation. Another parameter referred to destruction

1286

Ultrasound in Medicine and Biology

Volume 34, Number 8, 2008

Fig. 5. Postwall-filtered power as a function of acoustic pressure for a 5-MHz transducer with (a) 1-, (b) 3-, (c) 5- and (d) 10-cycle transmitted pulses.

percentage (DP) index was defined as 100% minus the SP index. When the insonation destruction pulse was off (i.e., no microbubble fragmentation), the estimated integrated imaging power was normalized to 1 and the DP index was 0%. All estimates of the integrated backscattered power were normalized to this value. EXPERIMENTAL RESULTS Results for different insonation parameters Figures 3 through 6 show the relationships between DP values and acoustic excitation pressure (from 0 to 1 MPa) for 1-, 2.25-, 5- and 7.5-MHz pulses, respectively. The hollow-circle symbols in Fig. 3a– d show the postwall-filtered normalized backscattered power integrated over the pulse index (slow time in Fig. 2d) for destruction pulses comprising 1, 3, 5 and 10 cycles. The insonation acoustic pressure was varied from 0 – 0.7 MPa for each pulse length. The error bars in the figures show one standard deviation estimated from 10 datasets recorded under each experimental condition. Figure 3 shows that the pressure thresholds for ⬎90% UCA destruction were 0.5, 0.5, 0.4 and 0.1 MPa for the 1-, 3-, 5- and 10-cycle pulses, respectively, indicating that the pressure threshold for microbubble destruction decreases with increasing pulse length. Figures 4 – 6 have the same format as Fig. 3. The results show

that the UCA destruction pressure threshold decreases with decreasing pulse frequency and increasing pulse length. More than 90% of the UCAs were destroyed at 0.1, 0.4, 0.6 and 0.7 MPa for 1-, 2.25-, 5- and 7.5-MHz 10 cycle–long destruction pulses, respectively. Model for predicting the destruction pressure threshold The results shown in Figs. 3– 6 indicate the presence of an exponential relationship between the DP estimates and insonation acoustic pressure. Therefore, we proposed the following exponential model (i.e., f(x)) to fit the discrete mean DP values weighted by the corresponding reciprocal of standard deviations as a function of pressure (i.e., weighted curve fitting):, f(x) ⫽ (1 ⫺ aexp ⫺bx) ⫻ 100%

(1)

where x is the pressure, a is the extrapolated DP value for an acoustic pressure of zero (i.e., the f(x) intercept point on the DP axis) and b is the intrinsic rate of decrease of the exponential model. All nonlinear fits were performed using a nonlinear least-squares algorithm (lsqnonlin function in MATLAB). The fitting results are shown as solid lines in Figs. 3– 6. The coefficient of determination (R2) between the fitted curves and the original DP estimates were all ⬎0.86. The high coefficients of determination indicate that our proposed model provides a good

Microbubbles destruction ● C.-K. YEH and S.-Y. SU

1287

Fig. 6. Postwall-filtered power as a function of acoustic pressure for a 7.5-MHz transducer with (a) 1-, (b) 3-, (c) 5- and (d) 10-cycle transmitted pulses.

description of the relationship between UCA destruction and the acoustic pressure. In the following analyses, the fitted model was interpolated to predict the degree of UCA destruction under different insonation conditions. The estimates of a and b obtained by fitting all the experimental data shown in Figs. 3– 6 using eqn (1) are listed in Table 2. From the results in Table 2, the b estimates generally increase with increasing pulse length and decrease with increasing the pulse frequency for each identical pulse length condition. In other words, the b estimate increasing is correlated with an increased probability of UCA fragmentation.

Based on the model fitting results, the acoustic pressure thresholds for DPs of 95%, 90%, 80% and 50% for different pulse lengths and frequencies are shown in Fig. 7a– d, respectively. Note that the error bars in these bar charts are the 95% confidence intervals on the thresholds taken from those fits at the indicated values of DP. In Fig. 7a, the pressure threshold of a DP of 95% decreases with increasing pulse length for each pulse frequency, and a lower pulse frequency results in a lower destruction pressure threshold for each pulse length. In Fig. 7a, the pressure thresholds for the pulse lengths of 3and 5-cycle in 2.25-MHz pulse is lower than that in

Table 2. The a, b and R2 estimates obtained from the exponential model with weighted curve fitting Pulse Frequency 1 MHz 2.25 MHz 5 MHz 7.5 MHz

Pulse Length

1-cycle

3-cycle

5-cycle

10-cycle

a b R2 a b R2 a b R2 a b R2

0.967 5.379 0.895 0.955 4.603 0.932 1.019 2.734 0.957 1.019 2.329 0.882

0.967 5.931 0.915 0.958 6.120 0.887 1.015 3.220 0.966 0.984 2.599 0.905

0.955 6.622 0.860 0.973 7.632 0.903 1.026 3.480 0.952 1.008 3.002 0.942

0.998 21.759 0.921 0.985 10.129 0.922 0.989 3.875 0.980 1.077 2.939 0.929

1288

Ultrasound in Medicine and Biology

Volume 34, Number 8, 2008

Fig. 7. Microbubbles destruction pressure thresholds, with destruction percentage (DP) of (a) 95%, (b) 90%, (c) 80% and (d) 50%.

1-MHz pulse. The possible reason is that the transient response of the microbubble will be much faster in the resonant case than off-resonance such that the radial oscillation of the microbubble is expected to be large, even for a short pulse (e.g., 3 to 5 cycles). Degree of microbubble destruction vs. MI The MI was derived by Apfel and Holland (1991) to predict the likelihood of transient cavitation in diagnostic ultrasound pressure fields. The formation and growth of microbubbles occurs during the negative (i.e., rarefactional) displacement of the pressure wave. The MI is defined as the peak rarefactional pressure derated by 0.3 dB/cm/MHz (Pr(0.3)) divided by the square root of the center frequency (fc) and thus is independent of both the pulse length and PRF. Therefore, the MI predicts that cavitation is less likely to occur at higher frequencies: MI ⫽

Pr(0.3)

兹fc

.

(2)

In this study, we used the measured value of peak negative pressure to calculate the mechanical index (MIe, where “e” indicates “experimental”). Figure 8a– d shows the relationship between MIe estimates and DP

values for pulse lengths of 1, 3, 5 and 10 cycles, respectively. Figure 8 shows that, as the DP decreases, the MIe estimate decreases for each combination of pulse frequency and pulse duration. Further, the theoretical data supporting the MI show that, at the threshold for inertial cavitation, the ratio given in eqn (2) is nearly a single, constant value. The results in Fig. 8a– d for different pulse lengths do not support this prediction and thus the results should not be interpreted as supporting the use of the MI as the microbubbles destruction threshold. Because the MIe estimates should be independent of the pulse frequency and corresponding pressure threshold for an identical DP condition, the degree of UCA destruction is not related to the MI estimate. DISCUSSION AND CONCLUSIONS When microbubbles pass through a finite width beam with a higher flow rate, the higher PRF of destruction pulse number is required to ensure that the bubbles within the sample volume were destroyed completely. However, the higher PRF value of destruction pulse will induce the primary and secondary radiation forces to affect the bubbles destruction. Smith et al. (2007) demonstrated that the relationship

Microbubbles destruction ● C.-K. YEH and S.-Y. SU

1289

Fig. 8. Experimental mechanical index (MIe) of microbubbles destruction thresholds with (a) 1-, (b) 3-, (c) 5- and (d) 10-cycle transmitted pulses.

between the PRF number of destruction pulse and rapid fragmentation threshold of echogenic liposomes is dependent. The lower-frequency pulses with higher pulse number may lower a acoustic pressure threshold. To further explore the degree of microbubble destruction with different PRF values of destruction pulse, we did another similar experiment with PRF values from 250 Hz–16 kHz in 2.25-MHz destruction pulse. Note that all experimental conditions were the same as

previous ones except the contrast agent was changed to SonoVue microbubbles (Bracco Diagnostics, Inc., Milan, Italy). The results between DP estimates and PRF are shown in Fig. 9. Figure 9a and b show the results with acoustic pressures of 2.25-MHz destruction pulse of 0.2 and 0.4 MPa, respectively. The ‘Œ’ and ‘●’ symbols denote the pulse length of 1- and 3-cycle, respectively. The results show that increasing pulse PRF number increases the degree of microbubble destruction.

Fig. 9. Degree of microbubble destruction with different values of PRF from 250 Hz to 16 kHz under the conditions of 2.25-MHz destruction pulse with acoustic pressures of (a) 0.2 and (b) 0.4 MPa.

1290

Ultrasound in Medicine and Biology

In this study, we have used an acoustic-based experimental method to demonstrate the relationship between ultrasound exposure conditions and the degree of microbubble destruction. Our proposed method represents a simple and easy way to determine the microbubble destruction threshold in laboratory experiments. The destruction threshold is determined based on a group of microbubbles within a sample volume (and not just a single bubble) and thus more closely replicates real clinical conditions. The experimental results presented here suggest that the acoustic excitation pressure, pulse center frequency and pulse length affect the probability of microbubble fragmentation: increasing the pressure and the pulse length and decreasing the pulse frequency are correlated with an increased probability of fragmentation. These conclusions are consistent with those drawn from optical observations (Bouakaz et al. 2005; Chomas et al. 2000, 2001b). Note that fewer microbubbles were observed to float up (because of buoyancy effect) and accumulate at the top of the tube in such experiment arrangement. To resolve the problem, the tube would be held vertically to eliminate the retention of large microbubbles at the top of the tube as a result of floatation. We used an integrated backscattered power estimate to represent the number of microbubbles that remain within the sample volume after applying an insonation destruction pulse. Theoretically, the intensity of the backscattered signal is proportional to the relative change in the microbubble concentration. Yamada et al. (2005) found that the power of the received signal in harmonic power Doppler imaging was proportional to the bubble concentration for a constant applied acoustic pressure. Schwarz et al. (1993) showed that the pulsedwave Doppler audio intensity is proportional to the concentration of Levovist over a limited range (i.e., for small concentrations) and, hence, that the estimated backscattered power is inversely proportional to the degree of UCA destruction. We have also used an exponential model with a weighted curve fitting technique to estimate the DP as a function of pressure. The high coefficients of determination (⬎ 0.86) for the fitting results in Figs. 3– 6 indicate that the model provides a good description of UCA destruction under different excitation pressures and, hence, the model can be used to predict the pressure threshold for UCA destruction. Figure 8 indicates that there is no such relationship between microbubble destruction and physical MI, which is consistent with the conclusions that Smith et al. (2007) drew from the microbubble destruction as detected by changes in video intensity. Some limitations such as background intensity and logarithmic compression will influence such video intensity measurements. However, these do not apply to the present study, which

Volume 34, Number 8, 2008

used RF backscattered signals to estimate the degree of microbubble destruction. Predicting the degree of microbubble destruction is useful in contrast-assisted perfusion imaging techniques. Wilkening et al. (2000) proposed a flow estimation model with fractional microbubble destruction based on UCA time–intensity curves. The method is based on continuous infusion of microbubbles resulting in a constant UCA concentration in the blood. By a series of fast high-energy destruction pulses, the UCA concentration is decreased until a steady state is reached in which one pulse destroys as much UCA as flows into the imaging plane during one interpulse interval. Our proposed acoustic-based experimental method can provide the information of degree of microbubbles destruction for making such a flow estimation model feasible. Another application is in high-resolution destruction/replenishment perfusion estimation systems. We previously constructed a high-frequency UCA destruction/replenishment imaging system that exhibited a spatial resolution of 160 ␮m in two dimensions (Yeh et al. 2004, 2007). That system used a 1-MHz focused transducer for microbubble destruction and a 25-MHz spherically-focused transducer for pulse/echo imaging. Because the fractional UCA destruction at 25 MHz under a specific pressure condition can be predicted using our proposed method, our previous system could be simplified to use a single high-frequency transducer. Of course, this would require modification to the theoretical destruction/replenishment model proposed by Wei et al. (1998). Other potential applications include the induction of microbubble cavitation and the use of sonoporation for drug delivery and gene therapy. Acknowledgements—We thank Mr. S.-Y. Lu for his assistance with the experiments, and the reviewers for their valuable comments. This work was supported by National Science Council under grant NSC 94-2320B-007-007.

REFERENCES Apfel RE, Holland CK. Gauging the likelihood of cavitation from short-pulse, low-duty cycle diagnostic ultrasound. Ultrasound Med Biol 1991;17:179 –185. Bouakaz A, Versluis M, de Jong N. High-speed optical observations of contrast agent destruction. Ultrasound Med Biol 2005;31:391–399. Bloch SH, Dayton PA, Ferrara KW. Targeted imaging using ultrasound contrast agents. IEEE Eng Med Biol Mag 2004a;23:18 –29. Bloch SH, Wan M, Dayton PA, Ferrara KW. Optical observation of lipid- and polymer-shelled ultrasound microbubble contrast agents. Appl Phys Lett 2004b;84:631– 633. Chang PH, Shung KK, Wu SJ, Levene HB. Second harmonic imaging and harmonic Doppler measurements with Albunex. IEEE Trans Ultrason Ferroelectr Freq Control 1995;42:1020 –1027. Chen WS, Matula TJ, Crum LA. The disappearance of ultrasound contrast bubbles: Observations of bubble dissolution and cavitation nucleation. Ultrasound Med Biol 2002;28:793– 803. Chomas JE, Dayton P, May D, Allen J, Klibanov A, Ferrara KW. Optical observation of contrast agent destruction. Appl Phys Lett 2000;77:1056 –1058.

Microbubbles destruction ● C.-K. YEH and S.-Y. SU Chomas JE, Dayton P, Allen J, Morgan K, Ferrera KW. Mechanisms of contrast agent destruction. IEEE Trans Ultrason Ferroelectr Freq Control 2001a;48:232–248. Chomas JE, Dayton P, May D, Ferrara K. Threshold of fragmentation for ultrasonic contrast agent. J Biomed Opt 2001b;6:141–150. Chomas JE, Pollard R, Wisner E, Ferrara K. Subharmonic phaseinversion for tumor perfusion estimation. Proc IEEE Ultrason Symp 2001c;2:1713–1716. Dayton PA, Morgan KE, Klibanov AL, Brandenburger G, Nightingale KR, Ferrara KW. A preliminary evaluation of the effects of primary and secondary radiation forces on acoustic contrast agents. IEEE Trans Ultrason Ferroelectr Freq Control 1997;44:1264 –1277. Forsberg F, Liu JB, Merton DA, Rawool NM, Johnson DK, Goldberg BB. Gray scale second harmonic imaging of acoustic emission signals improves detection of liver tumors in rabbits. J Ultrasound Med 2000;19:557–563. Forsberg F, Shi WT, Merritt RB, Dai Q, Solcova M, Goldberg BB. On the usefulness of the mechanical index displayed on clinical ultrasound scanners for predicting contrast microbubble destruction. J Ultrasound Med 2005;24:443– 450. Frinking P, Cespedes E, Kirkhorn J, Torp H, de Jong N. A new ultrasound contrast imaging approach bases on the combination of multiple imaging pulse and a separate release burst. IEEE Trans Ultrason Ferroelectr Freq Control 2001;48:643– 651. Goldberg B, Raichlen J, Forsberg F. Ultrasound Contrast Agents: Basic Principles and Clinical Applications, ed 2. London: Martin Dunitz, 2001. Hirokawa T, Nishikage T, Moroe T, Kajima M, Hayashi M, Naito T, Yamane S, Shiota H. Visualization of uveal perfusion by contrastenhanced harmonic ultrasonography at a low mechanical index: A pilot animal study. J Ultrasound Med 2002;21:299 –307. Klibanov AL, Ferrara KW, Hughes MS, Wible JH, Wojdyla JK, Dayton PA, Morgan KE, Brandenburger GH. Direct video-microscopic observation of the dynamic effects of medical ultrasound on ultrasound contrast microspheres. Invest Radiol 1998;33:863– 870. Lotsberg O, Hovem JM, Askum B. Experimental observation of subharmonic oscillations in Infoson bubbles. J Acoust Soc Am 1996; 99:1366 –1369. O’Brien WD. Ultrasound– biophysics mechanisms. Prog Biophys Mol Biol 2007;93:1–3:212–255. Porter TM, Smith DAB, Holland CK. Acoustic techniques for assessing the Optison destruction threshold. J Ultrasound Med 2006;25:1519 – 1529.

1291

Schwarz KQ, Bezante GP, Chen X, Schlief R. Quantitative echo contrast concentration measurement by Doppler sonography. Ultrasound Med Biol 1993;19:289 –297. Shi WT, Forsberg F, Hall AL, Chia RY, Liu JB, Miller S, Thomenius KE, Wheatley MA, Goldberg BB. Subharmonic imaging with microbubble contrast agents: Initial results. Ultrason Imag 1999;21: 79 –94. Shi WT, Forsberg F. Ultrasonic characterization of the nonlinear properties of contrast microbubbles. Ultrasound Med Biol 2000;26:93– 104. Simpson DH, Chien TC, Burn PN. Pulse inversion Doppler: A new method for detecting nonlinear echoes from microbubble contrast agents. IEEE Trans Ultrason Ferroelectr Freq Control 1999;46: 372–382. Smith DAB, Porter TM, Martinez J, Huang S, MacDonald RC, McPherson DD, Holland CK. Destruction thresholds of echogenic liposomes with clinical diagnostic ultrasound. Ultrasound Med Biol 2007;33:797– 809. Wei K, Jayaweera AR, Firoozan S, Linka A, Skyba DM, Kaul S. Quantification of myocardial blood flow with ultrasound-induced destruction of microbubbles administered as a constant venous infusion. Circulation 1998;97:473– 483. Wei K, Le E, Bin JP, Coggins M, Thorpe J, Kaul S. Quantification of renal blood flow with contrast-enhanced ultrasound. J Am Coll Cardiol 2001;37:1135–1140. Wilkening W, Postert T, Federlein J, Kono Y, Mattrey R, Ermert H. Ultrasonic assessment of perfusion conditions in the brain and in the liver. Proc IEEE Ultrason Symp 2000;2:1545–1548. Yamada S, Komuro K, Mikami T, Kudo N, Onozuka H, Goto K, Fujii S, Yamamoto K Kitabatake A. Novel quantitative assessment of myocardial perfusion by harmonic power Doppler imaging during myocardial contrast echocardiography. Heart 2005;91:183–188. Yeh CK, Yang MJ, Li PC. Contrast-specific ultrasonic flow measurements based on both input and output time intensities. Ultrasound Med Biol 2003;29:671– 678. Yeh CK, Ferrara W, Kruse DE. High-resolution functional vascular assessment with ultrasound. IEEE Trans Med Imag 2004;23:1263– 1274. Yeh CK, Lu SY, Chen YS. Microcirculation volumetric flow assessment using high-resolution, contrast-assisted images. IEEE Trans Ultrason Ferroelectr Freq Control 2008;55:74 – 83.