Professional Documents
Culture Documents
doi:10.1016/S0301-5629(03)00027-9
● Original Contribution
Abstract—In recent years, our knowledge of the behaviour of ultrasonic microbubble contrast agents has
improved substantially through in vitro experiments. However, there has been a tendency to use high concen-
trations of contrast agents in suspension, so that ultrasonic backscatter data are generated by a cloud of
microbubbles. Such experiments involve a variety of assumptions with validity that is open to question. In
addition, high concentrations of microbubbles cannot be used to understand the behaviour of individual
microbubble scatterers. This paper proposes a technique that minimises the number of assumptions that need to
be made to interpret in vitro experimental data. The basis of the technique is a dilute suspension of microbubbles
that makes single scattering events distinguishable. A commercial scanner was used to collect radio frequency
(RF) data from suspensions of two different contrast agents, Quantison™ and Definity. Backscatter data were
collected over a range of acoustic pressures. It was found that Definity provided a constant number of scattering
events per unit volume of suspension for almost all applied acoustic pressures. Quantison™ demonstrated an
increasing number of scattering events per unit volume with increasing acoustic pressure. Below 0.6 MPa,
Quantison™ scatterers were not individually detectable and provided levels of backscatter similar to those of a
blood-mimicking fluid, which suggests that Quantison™ microbubbles had almost linear scattering behaviour. At
acoustic pressures greater than 0.6 MPa, both agents appeared to provide echoes from free bubbles. The change
in the number of scatterers per unit volume with acoustic pressure cannot be demonstrated using high
concentrations of contrast agent. (E-mail: Vassilis.Sboros@ed.ac.uk) © 2003 World Federation for Ultrasound
in Medicine & Biology.
687
688 Ultrasound in Medicine and Biology Volume 29, Number 5, 2003
acoustic pressure has not been the focal point in most of centre frequency using an Ultramark 9 (UM9) ultrasonic
the above studies. scanner (Advanced Technology Laboratories, Bothell,
The above obstacles should not be a serious source WA). The nominal transmit power ranged from 2.24% to
of problems in the study of microbubble behaviour. The 100% of maximum output power and the focal distance
quantitative criterion “microbubble density less than that was 4.25 cm. This corresponded to a range of 0.27 MPa
required to cause shadowing” is often quoted in the to 1.52 MPa peak negative pressure at 3 cm from the
literature (Krishna and Newhouse 1997). This is usually probe, which was the centre of the ROI. The experiments
higher than 10 microbubbles per mL of suspension. were performed in triggered imaging mode. Single
Multiple scattering and bubble interactions have been frames were triggered, via the ECG lead, using a pulse
assumed insignificant, but there is evidence which sug- generator (Philips PM5705). Complete RF sector frames
gests that both phenomena might be underestimated were collected and images were reconstructed using a
(Soetanto and Chan 2000; Parlitz et al. 1999). But there moving average filter. The image analysis software writ-
is a more fundamental complication that is intrinsic to ten in the Department of Medical Physics (University of
the use of high microbubble concentrations. The distri- Edinburgh, UK) were used to calculate the RF backscat-
bution of microbubble responses is not a statistically ter power of subregions within an ROI. The subregions
normal distribution of responses, as has usually been were defined by thresholding the pixel intensity. In other
assumed. Quantison™ microbubbles, for example, have words, the average RF intensity of an ROI was calculated
a dual behaviour and scatter US either as free gas bubbles for those pixels that had higher pixel intensity value than
or as highly damped encapsulated bubbles (Sboros et al. a preset threshold. This allowed the calculation of aver-
2001). The scattering and absorption cross-sections of age intensities excluding background noise.
UCAs have been central to theoretical investigations and
their experimental assessment has always been a goal.
Even when the response of a region-of-interest (ROI) is The experimental protocol
measured and there are no further calculations to analyse The contrast agents used in this experiment were
the data, it is common to assume a normal distribution of Definity (DuPont Pharmaceutical Co., Waltham, MA),
microbubble behaviour in the discussion of the results. and Quantison™ (Quadrant Healthcare, Nottingham,
Studies with Quantison™ suspensions were limited in UK). The concentrations used were low enough to enable
quantifying the backscatter of the agent (Finking and de individual scatterers to be distinguished in the ultrasonic
Jong 1998; Frinking et al. 1999). Even for other contrast image and high enough to obtain the highest possible
microbubbles where the duality of Quantison™ is not number density to increase sample size and minimise the
present, it is difficult to justify the assumption of a variability of the measurements.
normal distribution of responses. To achieve this, both agents were diluted; 3 L of
Very diluted contrast suspensions are employed in Definity were introduced in a beaker with 100 mL of
the present study, to isolate single scattering events from sterile water and mixed for 1 min using a magnetic
two different contrast agents and to assess the usefulness stirrer. Of that suspension, 50 L were introduced into
of such an approach. There are two immediate advan- the tank, which contained 200 mL of sterile water, giving
tages of doing so: 1. the attenuation becomes negligible approximately 7.5 bubbles per mL (⬎ 1 m) of suspen-
and, therefore, does not need to be measured or calcu- sion. A period of 30 s mixing was allowed before the
lated and 2. single scatterers can be counted and a quan- acquisition of five triggered frames each separated by 1 s.
titative estimate of the scattering cross-section can be For Quantison™, 5 L were introduced into a beaker
obtained. The principal disadvantage of this approach is with 100 mL sterile water and, from that suspension, 250
the loss of signals that are at or below noise level and can L were introduced into the tank (94 bubbles larger than
only be registered collectively at high microbubble con- 1 m per mL of suspension). The experiment was re-
centrations. A range of acoustic pressures typical of peated at each of eight acoustic pressures, 5 times for
diagnostic scanning were used to assess their effect on Definity and 8 times for Quantison™.
contrast backscatter phenomena. No measures were taken to control the air content of
the suspensions. The decay of the either agent is not
significantly affected by air content (Sboros et al. 2000a).
MATERIALS AND METHODS
Both agents were also inspected in suspension under a
Technique phase microscope (Olympus IX50, Tokyo, Japan). All
The same experimental technique was used as in the microbubbles from both agent suspensions appeared to
previous communications (Sboros et al. 2000b, 2001). be separated from each other and no clustering was ob-
Suspensions of contrast agents were scanned at 3-MHz served.
Individual contrast microbubbles ● V. SBOROS et al. 689
Fig. 1. (a) Reconstructed image frame for Definity. The pixel intensity was calculated for the rectangle ROI (which was
also used in the rest of this paper). (b) Threshold 1 was employed and most of the pixels in the frame became blue. Those
pixels had higher amplitude than the threshold, and the analysis package would calculate the average pixel intensity for
the pixels inside the ROI. The threshold is the square root of the average pixel intensity of an identical ROI of a tank
filled with sterile water. Similarly in (c) and (d) Where thresholds 2 and 3 were used (square root of average pixel
intensity of two different are facts of the tank filled with sterile water). The images were the same frame at 0.58 MPa
peak negative pressure.
Data analysis following procedure. The tank was filled with sterile
water and was allowed to settle until no air bubbles or
Number of scatterers. The number of scatterers was
artefacts could be seen on the monitor. A frame was
counted manually in a specified ROI in each image. The
captured at each acoustic pressure. Three intensities from
ROI used was identical to that used previously (Sboros et
three different ROIs were calculated. The first one was
al. 2000b) and was 2.9 ⫾ 0.05 cm long and 1.4 ⫾ 0.05
from the ROI mentioned above and the square root of the
cm wide. The tip of a 0.2-mm needle hydrophone (Pre-
magnitude of the backscatter intensity was set to be
cision Acoustics, Dorchester, UK) was used as a point
threshold 1. The second and third ROIs were set over two
scatterer to assess the ultrasonic slice width. The scanner
different intensity backscattering artefacts appearing in
gain was adjusted to obtain an average pixel intensity
each frame. Their spatial peak intensities were still lower
from the tip of the hydrophone similar to the brightest
than those from individual microbubble backscatter in-
contrast microbubble. The slice width of the beam was
tensities. The respective square roots of those intensities
defined for the purposes of the present study to be equal
were threshold 2 and threshold 3. Both those artefacts
to the distance (across the slice width) at which the tip of
were air bubbles, visualised in the wall of the tank. They
the hydrophone was visible on the scanner’s monitor.
provided very low echoes because they were almost
This distance was found to be 1.50 ⫾ 0.05 cm. There-
outside the scan plane, but the echoes were significant
fore, the volume of the ROI was 6.2 ⫾ 0.3 cm. This
when compared with those from single scatterers. The
allowed an estimation of the number of microbubbles per
fourth threshold was set to zero to include background
unit volume imaged in the ROI to be compared with the
noise in the mean pixel intensity calculation (threshold
known concentration (7.5 or 94 microbubbles/mL).
4). After having set the threshold amplitude, all the
The measurement of backscatter. The average RF pixels that would be included in the average pixel inten-
intensity was calculated for the ROI. The analysis pack- sity calculation were displayed blue in the image. Figure
age installed on a Sun workstation was modified to allow 1b shows the pixels that were included in the calculation
the selection of a threshold pixel amplitude to calculate of the average pixel intensity using threshold 1. Figure 1c
the average RF intensity of the pixels that had higher and d shows the pixels that had amplitudes above thresh-
amplitude than that threshold. The RF intensity was old 2 and threshold 3. It is obvious that threshold 4 would
measured using four different thresholds produced by the create a completely blue image and, therefore, it is not
690 Ultrasound in Medicine and Biology Volume 29, Number 5, 2003
Fig. 2. Reconstructed frames for scatterer counting. The number of scatterers inside the rectangular ROI is stated in each
frame. The particular frames were chosen to have numbers of scatterers close to the average for all acquisitions at each
peak negative pressure. Definity showed a particularly constant number of scatterers at different acoustic pressures, but
Quantison™ showed an increasing number of scatterers with increasing acoustic pressure. Note that the number of
microbubbles in the suspension was 7.5 microbubbles/mL for Definity (⬃ 46 microbubbles in the ROI), and 94
microbubbles/mL for Quantison™ (⬃ 580 in the ROI). Note also that, at 0.79 MPa, there are two larger scatterers that
are probably air bubbles.
included in Fig. 1. The normalised backscatter was cal- onstrated approximately constant scatterer number at all
culated as in Sboros et al (2001). The four different acoustic pressures, apart from the lowest one. Quanti-
thresholds were compared by means of normalised son™ scatterers did not exceed the noise level of the two
backscatter. lowest pressures and demonstrated an increasing number
The ratio of the normalised backscatter (for a se- of scatterers with the increase of acoustic pressure.
lected threshold) to the number of scatterers in the ROI The calculation of the volume of the ROI allowed an
gives the normalised backscatter per scatterer. This cal- approximate calculation of the concentration of observed
culation was averaged over all frames captured at each
acoustic pressure. Because the bubble concentration was
very low, a small number of very large scatterers ap-
peared in some of the images, caused probably by im-
purities (usually air bubbles). Those frames could distort
the calculation of the normalised backscatter per bubble.
For this reason, those frames were not used in the cal-
culation of normalised backscatter.
RESULTS
Figure 2 shows reconstructed frames at two different
acoustic pressures for Definity and Quantison™. The
number of scatterers refers to the counts inside the rect-
angular ROI. The frames displayed in Fig. 2 were chosen Fig. 3. Number of scatterers counted in the ROI vs. peak
to have a number of scatterers close to the average for negative pressure. Definity provided constant counts at most
acoustic pressures and only the two lowest acoustic pressures
that acoustic pressure. had lower counts. Quantison™ showed an increase of the
The numbers of scatterers in the ROI at different number of scatterers with the increase of acoustic pressure that
acoustic pressures are illustrated in Fig. 3. Definity dem- settled at the two maximum acoustic pressures.
Individual contrast microbubbles ● V. SBOROS et al. 691
n nor REFERENCES
CA ⬇ c P⫺ (3)
n CA nor X Chang PH, Shung KK, Levene HB. Quantitative measurements of
second harmonic Doppler using ultrasound contrast agents. Ultra-
and, because nnor, nCA, nor, and Cx are constant, the sound Med Biol 1996;22:1205–1214.
D’Agostino L, Brennen CE. Acoustical absorption and scattering cross
average scattering cross-section of either contrast agents sections of spherical bubble clouds. J Acoust Soc Am 1988;84:
is proportional to the peak negative pressure. 2126–2134.
Definity provided an outlier to the linear trend at the Dayton P, Morgan K, Allietta M, et al. Simultaneous optical and
lowest acoustic pressure. The concentration of observer- acoustical observations of contrast agents. IEEE Ultrason Sympos
1997;2:1583–1591.
counted scatterers was very similar to the concentration Dayton PA, Morgan KE, Klibanov AL, Brandenburger GH, Ferrara
of bubbles (Fig. 4a), which suggests that any systematic KW. Optical and acoustical observations of the effects of ultra-
error in manually counting the scatterers was almost sound on contrast agents. IEEE Trans UFFC 1999;46:220–232.
Frinking PJA, de Jong N. Acoustic modeling of shell-encapsulated gas
negligible at most acoustic pressures. Furthermore, the bubbles. Ultrasound Med Biol 1998;24:523–533.
generally insignificant differences provided by the cal- Frinking PJA, de Jong N, Cespedes EI. Scattering properties of encap-
culations of normalised backscatter by the different sulated gas bubbles at high ultrasound pressures. J Acoust Soc Am
thresholds strongly suggests that the manual counting 1999;105:1989–1996.
Johnson R. Latest clinical developments with Quantison™. 2nd Tho-
of the scatterers was not underestimating their num- raxcentre European Symposium on Ultrasound Contrast Imaging,
ber. Therefore, the counted number of scatterers for Rotterdam, The Netherlands (Abst). January 1997.
the lowest acoustic pressure was correct and is prob- Krishna PD, Newhouse VL. Second harmonic characteristics of the
ultrasound contrast agents Albunex and FSO69. Ultrasound Med
ably due to the lower sensitivity of the system at that Biol 1997;23:453–459.
pressure. Krishna PD, Shankar PM, Newhouse VL. Subharmonic generation
It is difficult to draw conclusions for the behaviour from ultrasonic contrast agents. Phys Med Biol 1999;44:681–694.
Morgan KE, Dayton PA, Kruse DE, et al. Changes in the echoes from
of Definity at low acoustic pressures, however, because ultrasonic contrast agents with imaging parameters. IEEE Trans
only one point in the plot cannot be used to infer the UFFC 1998;45:1537–1548.
behaviour of the agent at low acoustic pressures. Parlitz U, Mettin R, Luther S, et al. Spatio-temporal dynamics of
acoustic cavitation bubble clouds. Phil Trans R Soc Lond A 1999;
SUMMARY 357:313–334.
Sboros V, Moran CM, Anderson T, McDicken WN. An in vitro
The behaviours of two different types of contrast comparison of ultrasonic contrast agents in solutions with varying
air levels. Ultrasound Med Biol 2000a;26:807–818.
agents have been explored by means of very diluted Sboros V, Moran CM, Anderson T, et al. Evaluation of an experimental
suspensions of contrast microbubbles. The high dilution system for the in vitro assessment of ultrasonic contrast agents.
allowed detection of single scattering events of the mi- Ultrasound Med Biol 2000b;26:105–111.
crobubbles, and it was found that the number of these Sboros V, Moran CM, Pye SD, McDicken WN. Contrast agent stabil-
ity: A continuous B-mode imaging approach. Ultrasound Med Biol
events for Quantison™ was dependent on acoustic pres- 2001;27:1367–1377.
sure. The scattering cross-section of the detected micro- Shi WT, Forsberg F. Ultrasonic characterization of the nonlinear proper-
bubbles was proportional to acoustic pressure. This was ties of contrast microbubbles. Ultrasound Med Biol 2000;26:93–104.
Shi WT, Forsberg F, Raichlen JS, Needleman L, Goldberg BB. Pres-
demonstrated for both agents, despite their different shell sure dependence of subharmonic signals from contrast micro-
composition and gas. There was also good evidence that, bubbles. Ultrasound Med Biol 1999;25:275–283.
above a certain acoustic pressure (around 0.6 MPa), free Soetanto K, Chan M. Fundamental studies of contrast images from
bubbles were released from both agents. Further studies different-sized microbubbles: Analytical and experimental studies.
Ultrasound Med Biol 2000;26:81–91.
of this type will be useful in accurately quantifying Wu J, Tong J. Experimental study of stability of a contrast agent in an
contrast agent backscatter. ultrasound field. Ultrasound Med Biol 1998;24:257–265.