- Email: [email protected]

doi:10.1016/j.ultrasmedbio.2006.07.031

● Technical Note FREQUENCY AND PRESSURE DEPENDENT ATTENUATION AND SCATTERING BY MICROBUBBLES MENG-XING TANG* and ROBERT J. ECKERSLEY† *Wolfson Medical Vision Laboratory, Department of Engineering Science, University of Oxford, Oxford, UK; and † Ultrasound Group, Imaging Sciences Department, Imperial College London, London, UK (Received 7 April 2006, revised 5 July 2006, in final form 20 July 2006)

Abstract—The aim of this study was to evaluate experimentally the degree of pressure dependence of attenuation and scattering by microbubbles at low acoustic pressures with an empirical nonlinear model. In addition, the pressure dependency over a range of frequencies (1 to 5 MHz) has been studied. A series of transmission and scattering measurements were made with the microbubble SonoVue™, using an automated system. Results show that, within the pressure range studied, attenuation as a result of the microbubble is pressure-dependent, whereas no such dependence of scattering was detectable. The pressure dependence of attenuation for SonoVue™ was found to be most significant at 1.5 MHz. The scattering is shown to be the highest at the lowest insonation frequency, around 1⬃1.25 MHz, and then decreases with frequency. (E-mail: [email protected]) © 2006 World Federation for Ultrasound in Medicine & Biology. Key Words: Ultrasound contrast agent, Attenuation, Scattering, Pressure dependency, Microbubbles, Multifrequency.

approaches, it is important to know whether the pressuredependent behaviour of microbubbles is significant over pressure variations within lower pressure ranges at different frequencies. Sboros (2002) reported experimental data of the pressure dependence of microbubble scattering for contrast agents Quantison and Definity, but only over a high-pressure range (269 kPa⬃1.5 MPa). At these pressures, microbubble disruption is, in general, significant and scattering is associated with bubble rupture and release of free bubbles. Frinking and de Jong (1998) showed experimentally that scattering coefficient of Quantison, a relatively hard-shelled microbubble contrast agent, is independent of acoustic pressure until the pressure reaches 200 kPa and beyond, when the scattering coefficient starts to increase significantly with acoustic pressure. On the other hand, models of microbubble dynamics do predict a dependence of scattering on acoustic pressure, the significance of this dependence being affected by shell rigidity (Chen et al. 2002). In their simulation, the hard-shelled model predicted a relative independence of scattering on pressure until high pressure (⬎300 kPa) is reached, whereas the relatively soft-shelled model predicted a strong dependence of scattering on pressure, even at pressure as low as 50 kPa. However, there currently are few experimental data on

INTRODUCTION Measurement and characterisation of attenuation and scattering by microbubble contrast agents have been under extensive study since their development (de Jong 1993; Leighton 1997; Marsh et al. 1998; Hoff 2001). In a number of studies, the attenuation and scattering of different contrast agents have been investigated with different concentrations and at different insonating frequencies (Marsh et al. 1998). However, there is another factor in microbubble scattering that has received less attention, namely the insonating acoustic pressure. Although it is, in general, accepted that microbubble behaviour changes from linear through nonlinear to disruption when insonating pressure increases, this pressure dependency is commonly deemed as being associated with large-scale pressure changes within a high-pressure range, e.g., from 200 kPa to 500 kPa or to 1 MPa. In a typical imaging situation, low acoustic pressure is commonly used for continuous microbubble imaging, but large spatial variation of acoustic pressure within the imaged plane exists as a result of attenuation and the acoustic field pattern. For current microbubble imaging Address correspondence to: Meng-Xing Tang, Wolfson Medical Vision Laboratory, Department of Engineering Science, Oxford University, OX1 3PJ, UK. E-mail: [email protected] 164

Bubble attenuation and scattering ● M.-X. TANG and R. J. ECKERSLEY

whether the pressure dependency of scattering is significant at low acoustic pressure for soft-shelled agents such as SonoVue™ and there are only a couple of experimental studies showing that the attenuation of SonoVue™ is significantly dependent on pressure even at a low pressure of ⬍64 kPa; (Chen et al. 2002). This paper is an extension of our previous work (Tang et al. 2005) and includes measurement of frequency dependence of attenuation and the frequency and pressure dependence of scattering. The pressure dependency of microbubble scattering and attenuation has important implications for image generation with current imaging techniques. With these techniques, it is always assumed that relative microbubble concentration in a target tissue can be obtained by measurement of the scattering coefficient. The underlying assumption is that scattering coefficient of microbubbles is independent of pressure. It is also assumed that attenuation as a result of microbubbles is independent of acoustic pressure, i.e., that microbubbles attenuate ultrasound linearly just as tissue. This, however, has been proven to be untrue. Because current techniques for imaging microbubbles make use of their nonlinear behaviour under ultrasound insonation, the microbubbles along the ultrasound transmission path between transducer and target attenuate the ultrasound transmission nonlinearly and contribute to the nonlinearity of the echoes. This process also can lead to imaging artifacts, especially in regions at depth (Eckersley et al. 2005). Therefore, a clear understanding of the pressure dependency for microbubble scattering and attenuation at multiple frequencies is vital for us to explore various sources of imaging artifacts and ways to correct them. The aim of this work is to observe the significance of the pressure and frequency dependence of attenuation and scattering by experimental measurements. METHODS Models of pressure-dependent attenuation and scattering When ultrasound transmits through a bulk volume of attenuators, the attenuation can be expressed as follows:

冉

S共x兲 ⫽ S0 exp ⫺

兰

x

x0

a共v兲dv

冊

(1)

where S0 is the initial pulse amplitude, S(x) is the attenuated pulse amplitude measured at location x, a is the attenuation coefficient and x0 is the beginning of the transmission path. Note that the ultrasound insonation is treated here as an impulse. This exponential attenuation model has been extensively used and some automatic

165

attenuation correction algorithms have been developed based on this model (Hughes et al. 1998). For nonlinear attenuators, such as microbubbles, the above model is no longer sufficient. The attenuation a due to nonlinear attenuators at location x has been previously modelled empirically by a first-order polynomial function of local acoustic pressure S(x) (Tang et al. 2005): a共x, S共x兲兲 ⫽ 共a0 ⫹ a1S共x兲兲c共x兲

(2)

where c(x) is the local concentration of microbubbles at location x and a0 and a1 are coefficients corresponding to baseline attenuation and pressure-dependent attenuation, respectively. Thus, the pressure-dependent attenuation can be modelled as follows:

冉

兰

S0 exp ⫺a0 S共x兲 ⫽

兰

1 ⫹ S0a1

x

x0

冉

x

x0

c共v兲dv

兰

c共u兲exp ⫺a0

x

x0

冊

冊

c共v兲dv du

(3)

This model has been validated experimentally at relatively low pressures (⬍219 kPa) for SonoVue™ with 3-MHz narrow-band pulses (Tang et al. 2005). As stated previously, microbubble scattering may be dependent on insonating acoustic pressure. Consequently, the microbubble scattering coefficient b can be modelled, similarly to that of attenuation, as a polynomial function of acoustic pressure S(x) b共x, S共x兲兲 ⫽ 共b0 ⫹ b1S共x兲兲c共x兲

(4)

where c(x) is the local concentration of microbubbles at location x and b0 and b1 are scattering coefficients, corresponding to the baseline scattering and pressure-dependent scattering, respectively. Experimental set-up The pressure-dependency of microbubble attenuation and scattering at multiple frequencies was investigated through in vitro experimental measurements. Both attenuation and scattering data were collected at multiple pressure amplitudes and multiple frequencies, and results were fitted to the proposed models to obtain the coefficients defined. There are two experimental set-ups, one for attenuation measurement and the other for scattering measurement. For attenuation measurement, the same experimental set-up was used as that described in Tang et al. (2005). SonoVue™ was diluted in saline (0.9%, w/v) and insonated using a laboratory ultrasound measurement system consisting of a pair of broadband 3.5-MHz single-element transducers (Videoscan V380, PanametricsNDT, Waltham, MA, USA). The transducers were aligned cofocally and submersed in a water bath main-

166

Ultrasound in Medicine and Biology

Fig. 1. An illustration of part of the insonating pulse series. Each burst contains 17 pulses, with frequencies ranging from 1 to 5 MHz. The pressure amplitude was increased by 10% between bursts. In total, 10 bursts were recorded for each experiment. The inset shows detail of a single pulse. (Note: for the sake of simplicity, the amplitude (y) axis and time (x) axis are scaled arbitrarily.)

tained at 37° C. A series of pulses with different amplitudes and frequencies were fired. An illustration of the pulse series is shown in Fig. 1. Each pulse consists of a Gaussian enveloped sinusoidal signal with four cycles. The centre frequencies of the pulses ranged from 1 to 5 MHz. There was a millisecond separation between any two consecutive sinusoidal pulses. The pressure ramp consists of 10 pressure levels corresponding to pressure ranges of 10.6⬃106 kPa (peak negative) for all frequencies. These relatively low acoustic pressures reduce the chances of both microbubble disruption and nonlinear propagation. Transmission measurements were made with microbubble concentration of 150, 225 and 300 L/1000 mL, respectively. The sample chambers used here have an acoustic path length of 4, 6 and 8 cm, respectively. These doses lie in the range similar to that used in a typical patient study if the clinical dose were evenly distributed throughout the blood pool. For every measurement with microbubbles, an identical microbubble-free control measurement was made. For scattering measurement, the two transducers were positioned at

Volume 33, Number 1, 2007

right angles to one another, focused on the microbubble suspension, immediately behind the acoustic window of the beaker. Because of the low signal-to-noise ratio in these scattering measurements, higher pressure is used at the higher frequencies. The pressure ramp consists of 10 pressure levels corresponding to pressure ranges of 10.6⬃106 kPa (peak-negative) at 1 MHz and 50.5⬃505 kPa at 5 MHz. For every scattering measurement with microbubbles, a microbubble-free control measurement was made with an acoustic mirror positioned at focus. The bubble scattering was then calculated as the total bubble scattered power normalised by the total power of the control measurement. The rest of the experimental conditions are the same as the attenuation measurement, except that only one sample chamber was used because acoustic path length is irrelevant in the scattering measurement. RESULTS Results of multifrequency attenuation Figure 2 shows the attenuation coefficients a0 and a1, which were obtained by fitting the model in eqn (3) with transmission measurements made. It can be seen that the baseline attenuation a0 initially increases in a linear fashion with frequency before reaching a plateau. The parameter a1 determines how much the attenuation depends on pressure. A significant peak can be seen at 1.5 MHz. Toward higher frequencies, a1 reduces to a low value and shows no dependency on frequency. These results show that, for SonoVue™, the attenuation is highly dependent on the insonating pressure around 1.5 MHz, suggesting this is the resonance frequency of the microbubble population. However, when higher frequency is used, the SonoVue™ microbubbles attenuate ultrasound with much less pressure dependency. Given a0 and a1, the attenuation of SonoVue™ at different frequencies can be conveniently synthesised for a given insonation pressure using eqn (2). Figure 3 shows exam-

Fig. 2. Attenuation coefficients a0 and a1 for SonoVue™ at multiple frequencies. (Note: the coefficient values shown in the figure correspond with diluted SonoVue™ suspension of 225 L/1 L.)

Bubble attenuation and scattering ● M.-X. TANG and R. J. ECKERSLEY

Fig. 3. Synthesised attenuation using a0 and a1 for SonoVue™ suspension of 225 L/1 L at different insonation pressures.

ples of such attenuation, with frequencies at a range of insonation pressures. Results of multifrequency scattering The original aim was to fit the model in eqn (4) to the scattering measurements to obtain the scattering coefficients b0 and b1. The measurements, however, show that the measured scattering signals do not show any significant dependence on insonation amplitude, as is shown in Fig. 4. Each curve in Fig. 4 corresponds to a specific insonating frequency and shows SonoVue™ microbubble scattering, normalised by control measurements, at 10 different insonating pressures. It can be seen that in each case, these can be reasonably modelled by a zero-sloped line, indicating no significant dependence of

Fig. 4. Average of 20 scattering measurements from SonoVue™ at different insonating pressures (pressure ramps 1 to 10, which correspond with pressure ranges 10.6⬃106 kPa (peak-negative) at 1 MHz and 50.5⬃505 kPa at 5 MHz ) and different frequencies.

167

Fig. 5. Scattering coefficient b0 for SonoVue™ at different frequencies (with b1 ⫽ 0). The error bar reflects variation of this coefficient as a result of different insonating pressures.

scattering on insonating pressure. As a consequence the parameter b1 in eqn (4) was set to zero and only an estimate of b0 obtained. In this case b0 is the average of the scattering data over the range of insonation pressures. Figure 5 shows the scattering coefficient b0 for SonoVue™ at multiple frequencies and the error bar reflects the variation caused by 10 different insonation pressures. It can be seen from Fig. 5 that, within the frequency range under study, the scattering from SonoVue™ is highest around 1.25 MHz and then generally decreases with frequency, which is in general agreement with previous studies (Marsh et al. 1998; Hoff 2001) DISCUSSIONS AND CONCLUSIONS In this paper, the pressure dependency of attenuation and scattering by SonoVue™ microbubble contrast agent has been studied at different frequencies. This pressure dependency, especially within a lower pressure range where microbubbles can be imaged constantly without being destroyed, has received little attention in the research community, although this potentially has significant implications in imaging results. The primary aims of this paper were, first, to evaluate the significance of this pressure dependence by experimental measurements and, second, to characterise the pressure-dependency at different insonating frequencies. The results show that the pressure dependency of the microbubble attenuation is significant. The model in eqn (3) is able to give a compact presentation of pressure-dependent attenuation using two parameters, i.e., a0, which determines baseline attenuation and a1, which determines how much the attenuation depends on pressure. The results show that a0 initially increases with frequency before reaching a plateau, and a1 peaks at

168

Ultrasound in Medicine and Biology

resonance and quickly decreases when the frequency is away from resonance. Given that a1 at its peak is 0.009 ⫾ 0.003 dB/cm/kPa while the corresponding a0 is 0.07 ⫾ 0.03 dB/cm (note that these values correspond to attenuation due to a dilute SonoVue™ suspension of 225 L/L), then for an insonation pressure of 100 kPa, the pressure-dependent attenuation is approximately an order of magnitude higher than the baseline attenuation. This pressure dependency is much less at frequencies away from resonance. For scattering, calculated as the total bubble scattered power normalised by the total power of control measurements, the results did not show any significant dependence on pressure. However, attenuation and scattering are not independent for a given bubble size and liquid. Therefore, the scattering of microbubbles is almost certainly pressure-dependent to some extent but, within the sensitivity of our experiment and within the low pressure range used in this study, we were unable to measure this. The pressure-dependent attenuation is easier to detect, as the attenuation builds up during the propagation of the pulses through the bulk suspension of microbubbles, whereas the scattering is a singular event, albeit the result of multiple bubbles within the focal region. This difference between the two phenomena means that we were not able, at these low pressures, to detect any pressure dependence in the scattering of microbubbles. Furthermore, the results show that, within the studied frequency range, scattering from SonoVue™ is highest around 1.25 MHz and then decreases significantly with increasing frequency up to 5 MHz. It should be noted that, at the higher pressure amplitudes used in the scattering experiments, some bubble disruption cannot be ruled out. However, the beaker used is relatively large (4-cm wide in the propagation direction of the ultrasound), and the focal region placed just

Volume 33, Number 1, 2007

inside the beaker was of the order of 1 mm3. The slow mixing of the bubble chamber would to some extent serve to introduce fresh bubbles into the focal zone. Furthermore, we argue that, if bubble destruction was a significant factor in these results, then the data obtained from the higher-amplitude experiments would differ from those of the low-pressure amplitude experiments, and this was not found to be the case (Fig. 4). Acknowledgements—Meng-Xing Tang is supported by the Engineering and Physical Sciences Research Council (EPSRC, grant number EP/ C536150/1). Robert J. Eckersley is funded by the EPSRC (grant number GR/S71224/01). The authors also thank MIAS-IRC (EPSRC grant number GR/N14248/01 and Medical Research Council grant number D2025/31) for their support. The authors also thank Dr. Eleanor Stride for her useful suggestions.

REFERENCES Chen Q, Zagzebski J, Wilson T, Stiles T. Pressure-dependent attenuation in ultrasound contrast agents. Ultrasound Med Biol 2002; 28(8):1041–1051. de Jong N. Acoustic properties of ultrasound contrast agents, PhD thesis. Erasmus University Rotterdam, The Netherlands, 1993. Eckersley RJ, Tang M-X. Nonlinear corruption of ultrasound transmission by microbubble contrast agents. Proceedings of the IEEE International Ultrasonics Symposium, September 18 –21, 2005, Rotterdam, The Netherlands. Frinking PJA, de Jong N. Acoustic modelling of shell-encapsulated gas bubbles. Ultrasound Med Biol 1998;24(4):523–533. Hoff L. Acoustic characterisation of contrast agents for medical ultrasound imaging, 1st ed. Dordrecht, The Netherlands: Kluwer Academic Publishers; 2001. Hughes DI, Duck FA. Automatic attenuation compensation for ultrasonic imaging. Ultrasound Med Biol 1997;23:651– 664. Leighton T. Acoustic bubble. San Diego, CA: Academic Press; 1997. Marsh JN, Hughes MS, Hall CS, et al. Frequency and concentration dependence of the backscatter coefficient of the ultrasound contrast agent Albunex. J Acoust Soc Am 1998;104(3):1654 –1666. Sboros V, MacDonald CA, Pye SD, et al. The dependence of ultrasound contrast agents backscatter on acoustic pressure: Theory versus experiment. Ultrasonics 2002;40(1– 8):579 –583. Tang M-X, Eckersley RJ, Noble JA. Pressure-dependent attenuation with microbubbles at low mechanical index. Ultrasound Med Biol 2005;31(3):377–384.