- Email: [email protected]
Single Bubble Acoustic Characterization and Stability Measurement of Adherent Microbubbles
Ultrasound in Med. & Biol., Vol. 39, No. 5, pp. 903–914, 2013 Copyright Ó 2013 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/$ - see front matter
http://dx.doi.org/10.1016/j.ultrasmedbio.2012.12.007
d
Original Contribution SINGLE BUBBLE ACOUSTIC CHARACTERIZATION AND STABILITY MEASUREMENT OF ADHERENT MICROBUBBLES JONATHAN CASEY,* CHARLES SENNOGA,* HELEN MULVANA,* JO V. HAJNAL,* MENG-XING TANG,y and ROBERT J. ECKERSLEY* * Imaging Sciences Department, Imperial College, London, UK; and y Department of Bioengineering, Imperial College, London, UK (Received 31 January 2012; revised 10 December 2012; in final form 11 December 2012)
Abstract—This article examines how the acoustic and stability characteristics of single lipid-shelled microbubbles (MBs) change as a result of adherence to a target surface. For individual adherent and non-adherent MBs, the backscattered echo from a narrowband 2-MHz, 90-kPa peak negative pressure interrogation pulse was obtained. These measurements were made in conjunction with an increasing amplitude broadband disruption pulse. It was found that, for the given driving frequency, adherence had little effect on the fundamental response of an MB. Examination of the second harmonic response indicated an increase of the resonance frequency for an adherent MB: resonance radius increasing of 0.3 ± 0.1 mm for an adherent MB. MB stability was seen to be closely related to MB resonance and gave further evidence of a change in the resonance frequency due to adherence. (E-mail: [email protected]) Ó 2013 World Federation for Ultrasound in Medicine & Biology. Key Words: Ultrasound, Microbubbles, Targeting, Adherence, Single bubbles, Resonance, Stability, Molecular imaging.
2004). By this attachment, the MB is brought to within tens of nanometers of the vasculature wall, which has been shown to affect the dynamic response of an MB to ultrasound (Caskey et al. 2006, 2007; Dollet et al. 2008; Garbin et al. 2007; Martynov et al. 2009; Ohl et al. 2009; Overvelde 2010; Sassaroli and Hynynen 2005). One of the main challenges associated with the use of targeted MBs for molecular imaging is how the signals from the attached MBs are differentiated from those of the free flowing or unattached MBs. The first approach typically used is that of clearance imaging. Unbound MBs are typically cleared from the circulatory system in approximately 10 min leaving the bound MBs for imaging (Lindner 2001). The main disadvantage with this technique is that although some MBs are retained at the target site, this number too decreases throughout the clearance period. Ideally, imaging should be performed at the time point of maximal MB retention and has directed many approaches to focus on image processing techniques to achieve this differentiation. Because of their compressible nature, MBs generate substantially more scattering than tissue and in particular generate more harmonic signal, either higher or subharmonics, even at low acoustic pressures (Frinking et al. 2010; Guidi et al. 2010). Such features allow MBs
INTRODUCTION Targeted microbubbles (MBs) have shown potential in allowing the use of ultrasound in a variety of new applications (Dayton 2007; Lindner 2004). By functionalizing the shell of the MBs to target specific markers expressed in the endothelium, the contrast agent can be used to highlight areas of interest and have been used in vivo to image inflammation (Lindner et al. 2000), angiogenesis (Willmann et al. 2008), atherosclerosis (Anderson et al. 2007) and thrombosis (Xie et al. 2005). Currently, targeted contrast agents for the molecular imaging of prostate cancer are undergoing initial human trials (Wijkstra et al. 2012). Their use as targeted drug delivery vehicles is also under investigation (Pitt 2004). These targeting ligands—be they antibodies, proteins or peptides dependent on the pathology to be targeted—are situated on the surface architecture of the MB, either on the shell itself or more commonly at the end of a spacer, typically a poly-ethylene glycol (PEG) spacer (Borden et al.
Address correspondence to: Robert J. Eckersley, Biomedical Engineering Department, Division of Imaging Sciences, King’s College London, St. Thomas’ Hospital, London, United Kingdom. E-mail: [email protected] 903
904
Ultrasound in Medicine and Biology
to be distinguished from tissue by image subtraction (Zhao et al. 2007) or examination of the harmonic signals generated. The differentiation between attached and unattached bubbles is then conducted by temporal filtering. An attached bubble will appear as a stationary scatterer, whereas an unattached bubble will appear as a moving scatterer; therefore, stationary bubbles can be detected by applying a low-pass filter to remove bubbles in motion. Several authors have successfully demonstrated variations of this technique in vitro (Needles et al. 2009; Zhao 2006; Zhao et al. 2007); however, in vivo a number of additional complications, such as body or tissue movement, make these techniques more difficult. A more efficient and selective imaging strategy could be developed if the signals generated by attached MBs showed unique acoustic characteristics that could then be identified specifically. The acoustic characteristics of MBs under a variety of conditions must be determined precisely so that specific imaging protocols can be devised. It is known that the local conditions can have a significant effect on MB response. Increasing values of wall rigidity decrease MB resonance frequency. When an elastic wall is modeled in these cases, it can have the effect of increasing or decreasing the resonance frequency depending on the specific material properties of both the wall and the surrounding medium. In most physiologically analogous situations, this has been shown to increase MB resonance frequency (Caskey et al. 2007; Doinikov et al. 2011; Martynov et al. 2009; Qin and Ferrara 2007). Patil et al. (2008) conducted a 3-D finite element examination of an MB adherent to a rigid wall in comparison with a free, unconstrained MB. In addition to showing many of the shape fluctuations as seen in highspeed imaging experiments (Dollet et al. 2008; Vos et al. 2008), the model also predicted a shift in resonance frequency as a result of adherence, having the effect of lowering the resonance frequency and suppressing some of the second harmonic signal generation. These observations have been supplemented by a high-speed imaging study (Overvelde et al. 2011); however, acoustic verification of these results has not yet been presented. Furthermore, MB destruction thresholds and unstable mode oscillations have been shown to be affected by the presence and characteristics of nearby boundaries; however, the presence and effects of binding on MB stability are not fully understood. Couture et al. (2009) demonstrated that there is also a difference in the dissolution time of adherent MBs compared with non-adherent MBs after ultrasound exposure. Adherent MBs were shown to dissolve at a significantly faster rate than their nonadherent counterparts. This finding could provide a differentiation technique based on the disruption of MBs; however, this would require high temporal resolution to be achieved.
Volume 39, Number 5, 2013
Any difference, such as changes to resonance frequency, detected between adherent and non-adherent MB acoustic signatures is expected to be subtle and size dependent. As such, bulk acoustic measurements—for which scattering and attenuation measurements are performed on bulk suspensions of MBs—are unlikely to elucidate the important details. Consequently, the acoustics characteristics of single MBs in varying conditions will be investigated in this work. The acoustic characterization of individual MBs has been implemented by a number of authors (Guidi et al. 2010; Sboros et al. 2003, 2005; Guidi et al. 2006; Vos et al. 2009) who have highlighted phenomena including acoustically induced deflation, compression and expansion-only behavior and changes in bubble stability when confined in small vessels. The objective of this study is to perform an acoustic investigation of the effect that binding an MB to a surface has in comparison with an unbound MB in otherwise similar conditions (i.e., near a boundary yet unattached). Specific attention will be given to both the resonance characteristics of the MBs and changes in bubble response to repeated insonation. MATERIALS AND METHODS Experimental setup The equipment (Fig. 1) consisted of a pair of matched transducers (V380, Panametrics, Olympus Industrial, Southend-on-Sea, UK) (4 MHz center frequency, 26 dB bandwidth 86%, focal length of 75 mm) focused to a central point in conjunction with a 403 water-immersible objective (LUMPlanFL N40xW, Olympus Medical, Southend-on-Sea, UK). Positioned at the focus was a capillary fibre (200 mm diameter; RC55 8/200, Membrana, Wuppertal, Germany) through which the MB suspension was flowed. Sham experiments, conducted without the presence of MBs, showed that the fiber did not scatter sound at a detectable level using the current setup and was therefore considered both acoustically and optically transparent. The fiber was mounted on a three-axis translational stage (562-XYZ ULTRAlignÔ Precision Linear Stage, Newport Corporation, Irvine, CA, USA) for accurate positioning. Note that to minimize acoustic interference, particular care must be taken over the alignment of the transducers and optics. The transducers were positioned approximately 10 degrees off axis from the objective. Similarly, the angles between the two transducers were unequal (positioned approximately 20 and 45 degrees from the normal for receiving and transmitting transducers, respectively) to minimize direct reflections. An additional complication was that, because of the required proximity of the objective to the capillary fiber (limited by the objective working distance of 3.3 mm), the presence of a scatterer in the fiber caused complex interference in the received signals. To remedy this, once an optical image had been
Characterization of adherent microbubbles d J. CASEY et al.
905
Fig. 1. Schematic of the experimental setup.
acquired to size the MB, the objective was removed from the acoustic field before taking the acoustic measurements. The transmitted signal was generated in an arbitrary waveform generator (Sony Tektronix AWG2021) and amplified through a power amplifier (2100L; E&I, Rochester, NY, USA). The scattered signal was amplified by a pulser/receiver operated in receive mode (Panametrics-NDT 5800) and displayed via a digital oscilloscope (Sony Tektronix TDS7154, Tektronix UK Ltd., Bracknell, UK). The image from the objective was directed to a digital camera (Canon ProShot G5) through a 45-degree mirror and focusing lens for subsequent sizing. Optical calibration was performed using latex sizing beads (3, 5 and 10 mm; Beckman Coulter, Brea, CA, USA). The beads were optically imaged and sized based on image intensity to determine their radius. To determine the robustness of the sizing technique, multiple images of populations of the three different beads were examined. This examination resulted in an average error of 0.08 mm and a maximum error of 0.14 mm independent of particle size. This approach is based on the work of Helfield et al. (2012). Final optical calibration was found to be 20 pixels/mm. The entire system was controlled via a desktop computer and an internally developed MATLAB program (Mathworks, Cambridge, UK). The gas saturation of the solution has been shown to affect the MB stability and the reproducibility of measurements (Mulvana et al. 2012). In accordance with these findings, filtered, gas-saturated water was used for all experiments. The water was filtered via reverse osmosis (euRO 20; Triple Red Ltd., Long Crendon, Buckinghamshire, UK) and was considered deionized. All experiments were performed at 20 C. Microbubble preparation Microbubbles were prepared by sonication (Misonix Sonicator 3000; settings: 21 kHz, 165 W, 30 s) of
an octo-fluoropropane–saturated aqueous suspension of distearoyl-phosphatidylcholine, distearoyl-phosphatidylethanolamine-PEG2000-biotin (DSPE-PEG2000-Biotin) and poly(ethylene glycol)-monostearate (PEG40-stearate). Non-targeted MBs (control) were prepared similarly, substituting DSPE-PEG2000-Biotin with methoxy-poly (ethylene glycol)2000 distearoyl-phosphatidylethanolamine (DSPE-mPEG2000). MB size distribution and concentration were measured to have a mean size diameter of 2.4 6 0.4 mm and a number concentration of 1.2 3 109 MBs/mL using optical microscopy (Sennoga et al. 2010). To conduct these experiments, it was vital to ensure that only a single bubble was present at the acoustic focus. To achieve this requirement, the MB solution was diluted by approximately 1:100,000 with some variation introduced because of bubble buoyancy, sampling errors and variations in the initial concentration of the MB solution. The transducer focal region was measured to be approximately 2.4 mm in diameter. The capillary fiber was checked optically to a distance of .1.5 mm either side of the target bubble to verify that no other bubbles were present. Upon injection, MBs rise to the upper inner surface of the capillary fibre, which as stated in the introduction brings with it some profound changes in bubble response. Because this change will affect both the adherent and nonadherent MBs, it was deemed acceptable and favorable, allowing direct comparison of bound and unbound bubble without any changes in local conditions or geometry.
Tube preparation Capillary fibers were cleaned with sterile phosphatebuffered saline (Sigma-Aldrich, Dorset, UK) and flooded with unlabeled streptavidin (Invitrogen Life
906
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
Fig. 2. Pulse definition; 2-MHz, 10-cycle narrowband interrogation pulse (top left); broadband 3.5-MHz, two-cycle destruction pulse (top right). Whole-pulse sequence showing the first five interrogation pulses [bottom (a)], increasing ramp of destruction pulses interleaved with the interrogation pulses [bottom (b)].
Technologies, Paisley, UK) solution at a concentration of 0.25 mg/mL before incubation overnight at 4 C. Acoustic pulse characteristics Accurate and detailed transducer characterization was essential to the success of this investigation. To obtain the frequency response spectrum, a plane reflector was positioned at the focus of the receive transducer. Operating the transducer via the pulser/receiver, the transducer was driven with an impulse and the subsequent response was recorded (details of bandwidth and center frequency given in the Experimental Setup section). This response is the frequency response after passing through both the transducers twice (transmitting and receiving). Assuming transducer reciprocity, the response for travel in one direction was taken as the square root of the full response. Spatial and pressure characterization of the ultrasound field was conducted using a calibrated 1-
mm–diameter needle hydrophone (HPM1/1; Precision Acoustics, Dorset UK). Each MB was insonated with a pulse sequence split into two distinct phases. Phase one (Fig. 2a) consisted of five, 2-MHz, 10-cycle pulses at a peak negative pressure (PNP) at the focus of 90 kPa. Phase two (Fig. 2b) consisted of alternating two-cycle, 3.5-MHz, center-frequency broadband pulses and the same 2-MHz narrowband pulse as in phase one for 10 repetitions (20 pulses in total). After each repetition, the broadband pulse amplitude was increased in increments of approximately 45 kPa from 50–450 kPa PNP. All pulses are sent at a pulse repetition frequency of 100 Hz. The rationale for this pulse sequence was twofold. The first narrowband pulse will give the bubble’s acoustic behavior at the bubble size imaged by optical microscope. The repeated pulses will then show how the bubble responds to repeated exposure. It is
Characterization of adherent microbubbles d J. CASEY et al.
907
Fig. 3. Typical acoustic responses for MBs at resonance (left) and away from resonance (right). Top panes show sizing photos, and the middle and bottom panes show the bubble response in the time and frequency domains, respectively. Scale bar 5 5 mm. MBs 5 microbubbles.
understood that MB stability is significantly affected by resonance (Chomas et al. 2001); it was intended that the first five pulses should provide detailed information on the acoustic and stability characteristics of MBs at or near resonance for a given narrowband excitation. Because the size distribution of bubbles examined is dependent on their initial size distribution and longevity after extreme dilution, the technique is biased toward larger bubbles. For bubbles with radii away from that of the resonance size, it has been shown that little disruption occurs (Guidi et al. 2010). Therefore, in an effort to maximize the data obtainable from MBs away from resonance, broadband pulses of increasing amplitude were used.
Data analysis The data for each bubble was split into single pulses, corrected for amplifier gain, zero offset and windowed using a hamming window. The data were then corrected for the frequency response characteristics of the transducer.
Figure 3 shows both the time and frequency domain responses recorded for two adherent MBs of different sizes. The left panels show the response of an MB near resonance. Note that for a 2-MHz insonation pulse, the resonance radius for a lipid-encapsulated bubble is predicted to be 1.8 mm (Morgan et al. 2000, van der Meer et al. 2004). This near resonance response is clearly demonstrated by the strong second harmonic component in the frequency spectrum. The right panels show the response of a bubble that is greater than the resonance size for a 2-MHz insonation pulse. Fundamental scattering can be seen to be increased, and the generation of the second harmonic signal is diminished. In the results that follow, scattering power is defined as the total area under the frequency spectrum curve normalized by the noise floor (see eqn [1]). Similarly, the scattering power of the separate harmonics was taken as the band-pass area under the harmonic peak (0.5 MHz either side of the peak frequency). By tracking the changes in scattering power of the various harmonics as a function of bubble size, a detailed picture of MB acoustic characteristics can be determined.
908
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
"Pfhigh Scattering PowerðdBÞ 5 20 log10
where fhigh and flow refer to the upper and lower bounds of the frequency range, corresponding to the peak frequency (60.5 MHz). To evaluate the change in bubble response to repeated insonation, the total scattering power was tracked as a function of interrogation pulse number. Figure 4 shows the normalized scattered power against the pulse number for three bubbles undergoing different changes in response. Scattered power is normalized against the response from the first pulse. Three distinct modes of MB response are visible. The green trace shows an MB that does not alter in response power over the entire pulse sequence as demonstrated by a normalized power remaining at 1. The blue trace shows an MB that undergoes a reduction in scattering response immediately after the first insonation pulse. The red trace shows an MB that does not exhibit a change until after the first five interrogation pulses. In reference to Figure 2, this indicates that the increasing ramp of broadband disruption pulses is responsible for the change in response. To parameterize this change in response, the area under each response curve was found and normalized against a constant response of 1 (i.e., the response of an MB exhibiting no change); this gives a single value per MB. A value of 1 is equivalent to an MB with no change in response. A value of less than 1 shows a decrease in response. When plotted against initial MB radius, the
flow signal amplitude spectrumðf Þ Pfhigh flow noise amplitude spectrumðf Þ
# (1)
size-dependent change in response was obtained. This parameter will henceforth be referred to as the ‘‘stability index.’’ Individual data points were subject to a large degree of variation, the cause of which will be discussed later. To extract trends within the data, the mean and standard deviation of a five-point moving window was calculated for each data point in both the sizing (x-axis) and parameter (y-axis) directions. Note that the end data points were subject to smaller widow sizes; the first and last data points used a three-point window, and the second and penultimate points used a four-point window. On subsequent figures using this windowing, mean data are displayed as a line with shaded regions representing the standard deviation of the mean. Raw data are displayed as discrete points. To show the statistical significance between the comparison trends, the data are binned into 0.2-mm bins (conservatively above the maximum sizing error). This process results in bins ranging in size up to 12 data sets per bin depending on the spread of MB sizes interrogated. The data are then tested using a two-tailed unpaired t-test to determine whether the two sets of data are significantly different. The significance level is set to p . 0.05, and the regions of significant difference are highlighted in the relevant figures. RESULTS
Fig. 4. Normalized total scattering as a function of interrogation pulse number to show bubble response as a function of pulse repetition. The green line shows a bubble with no change in response over the pulse sequence. The blue line shows a bubble exhibiting a change in response subsequent to the first insonation. The red line shows a bubble exhibiting a change in response as the destructive broadband pulses are applied (post pulse 5).
Scattered acoustic power The following data present the results from all four experimental conditions tested: (i) untargeted MBs, uncoated capillary fiber (UTUC); (ii) untargeted MBs, coated capillary fiber (UTC); (iii) targeted MBs, uncoated fiber (TUC); and (iv) targeted MBs, coated fiber (TC). The number of data points for each of the test cases were: TC, n 5 41; UTUC, n 5 44; TUC, n 5 49; and UTC, n 5 45. By comparison of these cases, the effect that the functionalization of both the bubbles and the capillary fibre has on MB dynamic response can be elucidated. For display purposes, each case is shown individually, and for comparative analysis the mean trend lines are superimposed on each other. Figure 5 shows the fundamental scattering results. In general, good agreement can be observed between all testing regimes, especially through the resonance region where the greatest variation was anticipated (p . 0.05
Characterization of adherent microbubbles d J. CASEY et al.
909
Fig. 5. Fundamental scattering as a function of bubble radius for the four testing regimens, showing the three nonadherent or control cases of untargeted–uncoated (top left), targeted–uncoated (top right), untargeted–coated (middle left) and the adherent or targeted–coated (middle right) regimen. Mean trend lines are superimposed to show the comparison (bottom); p . 0.05 for the entire radius range.
for the entire range radius). Figure 6 displays the second harmonic scattering data. Good agreement is observed among the three non-adherent cases (i.e., UTUC, TUC and UTC; p . 0.05) both in terms of resonance radii and scattering power. The slightly lower peak values for the UTUC case could be explained by the limited number of data points within this region (Fig. 5, top left). The adherent MBs are statistically different (p , 0.05) from all three non-adherent cases in the radius range of 1.8–2.0 mm. To establish the radii associated with peak second harmonic generation, the data were fitted with a polynomial curve. Figure 7 shows this curve fitting for the adherent MBs. The radius at which maximum second harmonic generation occurs is indicated by the solid line (R 5 1.86 mm). To provide confidence bounds for this peak, the mean residual was subtracted from the peak second harmonic scattering power; this provided a range of MB radii of 1.75–1.97 mm. These data have been collated for all testing regimens in Table 1. Given the similarities among the three unbound cases, the following sections consist of comparisons of the adherent (TC) and the non-adherent (TUC) MBs to identify differences among otherwise identical bubbles with or without target receptors. Figure 8 shows the MB scattering power as a function of MB radius for fundamental scattering (top) and second harmonic scattering (bottom). The fundamental scattering is minimal below the resonance radius. Through the resonance radius for a 2-MHz insonation pulse (1.8 mm based on current theory for phospholipid shelled MBs; Morgan et al. 2000; van der Meer et al. 2004), there is an increase in the fundamental scattering until a radius of 2 mm, at which point the rate of increase
levels out. This trend is consistent in both the adherent and non-adherent MBs. There is no statistical difference between the trends. The second harmonic scattering power shows a maximum at approximately where one would expect the resonance radius to occur. Peak maxima were extracted (Table 1) and showed an adherent MB resonance radius of 1.86 mm and a non-adherent resonance radius of 1.60 mm; the two data sets have been shown to be statistically different in the range of 1.8–2.0 mm (p , 0.05). Away from resonance effects (.2.5 mm), the response characteristics between the two bubble types are in agreement with diminished second harmonic generation. MB response to repeated insonation Figure 9 (top) shows the bubble stability index after the first five insonation pulses. A stability index of 1 refers to a bubble whose response is invariant with repeated insonation. A stability index less than 1 indicates that the MB response decreases with subsequent insonation pulses. A number of bubbles can also be seen to have a stability index of greater than 1, indicating an increase in acoustic response to successive insonations; the possible reasons for this are explained in the Discussion. MBs near the peak second harmonic generation as seen in Figure 8 for both adherent and non-adherent MBs are subject to the most disruption. Upon exposure to the first five interrogation pulses, non-adherent MBs first experience a decrease in response at radii ,1.8 mm; this is in comparison with adherent MBs that experience this response at a larger initial radius (,2.2 mm; p , 0.05 in the range of 1.8–2.0 mm). The size below which the MB response is seen to decrease will be called
910
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
Fig. 6. Second harmonic scattering as a function of bubble radius for the four testing regimens showing the three nonadherent or control cases of untargeted–uncoated (top left), targeted–uncoated (top right), Untargeted–coated (middle left) and the adherent or targeted–coated (middle right) regimen. Mean trend lines are superimposed to show the comparison (bottom). All three non-adherent regimes have a p . 0.05 throughout the radius range examined. The adherent case differs (p , 0.05) from the non-adherent cases in the radius range of 1.8–2.0 mm.
the ‘‘disruption threshold.’’ As with the second harmonic generation, for MBs with a radius greater than 2.5 mm, the MBs can be seen to be essentially invariant with pulse repetition for both adherent and non-adherent MBs (p . 0.05). Figure 9 (bottom) shows the stability index after the bubble has been exposed to the whole pulse sequence (i.e., insonated with the interrogation pulses and the increasing pressure ramp of broadband insonation); in general, it shows the same trend as Figure 9 (top), showing the size dependence on the stability index. Note that the MB disruption threshold for both MB cases has increased, from approximately 1.8–2.0 mm for nonadherent MBs and from 2.2–2.4 mm for adherent MBs. Furthermore, the levels of disruption experienced in
both cases are dramatically increased, and it follows that the curves are significantly different over a wider range (p , 0.05 in the range 1.8–2.4 mm). DISCUSSION Experimental errors and variability The data presented shows a large degree of variability, even for similarly sized bubbles, the cause of which may be attributed to several factors. Although bubble response is strongly dictated by initial bubble radius, the bubble’s structural properties have also been shown to have a significant effects, specifically initial surface tension resulting in differences in both amplitude and shape of oscillation (e.g., compression-expansion– only behavior, bubble destruction and fragmentation; Marmottant et al. 2005; Overvelde et al. 2010). During insonation, the MBs are all exposed to a primary radiation force resulting in bubble motion. Given a sufficiently large motion, this could move the MBs out of the focal region of the transducers, giving rise to misleading results. To certify that this would not be a problem, a number of bubbles were insonated with the full pulse as described earlier, but with the microscope Table 1. Resonance radii as derived from curve fitting to second harmonic generation
Fig. 7. Example of curve fitting to the second harmonic scattering data of adherent MBs for resonance detection. Curve maximum occurs at R 5 1.86 mm, indicated by the solid black line. Dashed lines show the confidence interval for peak detection. MBs 5 microbubbles.
Testing regime
Resonance radius (mm)
Confidence interval (6 mm)
Untargeted–uncoated Targeted–uncoated Untargeted–coated Targeted–coated
1.59 1.60 1.60 1.86
0.11 0.09 0.10 0.11
Characterization of adherent microbubbles d J. CASEY et al.
911
Fig. 8. Scattering power as a function of MB radius for fundamental scattering (top) and second harmonic scattering (bottom). Comparison between adherent MBs (dashed blue line) and non-adherent (targeted MBs in an uncoated capillary fibre, solid red line). MBs 5 microbubbles.
objective still in place to directly observe bubble motion. Maximal bubble motion was observed to be of the order of less than 10 mm. Another possible source of error comes from the presence of nearby bubbles in the acoustic field. Although this problem is minimized by the optical verification that the capillary fiber is clear of either side of the target bubble, only the upper surface of the fiber is scanned for time saving purposes and because any MBs should rise through buoyancy; therefore, it does not mitigate against bubbles being entrained elsewhere on or in the fiber, either trapped by particulate matter or forming because the whole system is immersed in gas saturated water. Typically, any large bubbles of this kind are detectable from the magnitude and shape of the response and discounted; however, for smaller bubbles these responses could be overlooked easily, leading to skewed data. Scattered acoustic response All the bubbles examined displayed similar fundamental response to the 2-MHz insonation pulse, whether they were adherent or non-adherent, indicating that under these conditions the presence of binding does not change the fundamental response of an MB. However, examination of the second harmonic scattering exhibits a difference. The shift in resonance radius between the two cases (1.86 mm for the adherent MB compared with 1.60 mm for the non-adherent MB) indicates an increase in resonance frequency for the adherent MBs. The difference between the resonance radii of the adherent (TC) compared with the three non-adherent MB cases indicates that adhesion causes the shift in resonance radius (and therefore frequency) and is not simply an artifact
of the functionalization of either the capillary fiber or the MB. Although the contribution of the second harmonic can be seen to have a dramatic effect on the level of total scattering, especially around resonance, it does not provide a sufficient difference to delineate the adherent from the non-adherent MBs on its own. This finding in itself is important when trying to implement differentiation protocols in more complex scenarios, implying that detailed frequency content is required rather than just total backscatter in the differentiation of adherent and free MBs. After comparing these results to the findings of Patil et al. (2008) and Overvelde et al. (2011), a number of differences present themselves. All three studies (this study included) show differences in the acoustic characteristics for adherent MBs; however, the way this difference is demonstrated is different in each case. The Overvelde study would appear to be most comparable with the present study, because of similar frequency ranges, bubble type and applied pressures. By insonating both adherent and non-adherent MBs of similar size (radius z 2.1 mm) through a range of frequencies and recording their oscillations via a high speed camera, the frequency of maximum response was found. This frequency of maximal response was then nondimensionalized by normalizing against the resonance frequency of an uncoated free MB of the same radius. The uncoated free resonance frequency given by: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 1 1 2s 2s w w f0uncoated 5 3g P0 1 2 (2) R0 R0 2p rR20 where r is the liquid density (998 kg/m3), R0 is the initial MB radius, g is the polytropic gas exponent (1.07), P0 is
912
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
Fig. 9. Stability index for the first five pulses of pulse sequence (top) showing a comparison between adherent MBs (dashed blue line) and non-adherent MBs (solid red line); p , 0.05 in the radius range of 1.8–2.0 mm. Stability index after exposure to the whole pulse sequence (bottom) is shown in a comparison between adherent MBs (dashed blue line) and non-adherent MBs (solid red line); p , 0.05 in the radius range of 1.8–2.4 mm. MBs 5 microbubbles.
the ambient pressure (101 kPa) and sw is the interfacial tension coefficient (0.072 N/m; Doinikov et al. 2009). Given the large discrepancies between the findings of this study and the Overvelde study for seemingly similar acoustic conditions, one must then assume that there is a difference in other experimental conditions or in the way results are analyzed that causes this discrepancy. There are a number of factors that could influence the MB response in these studies. For example, any differences in MB shell chemistry effectively change the MB visco-elastic properties between the two studies. This has the effect of modifying the shape and position of the resonance frequency curve. In addition, it has been postulated that both shell elasticity and viscosity have a size-dependent nature (Sboros et al. 2006; Tu et al. 2009, 2011), further skewing the resonance frequency curves. As an aside to shell chemistry, the binding mechanisms used in the two studies are also different. A biotin–streptavidin ligand pair in this study compared with a fluorescein–anti-fluorescein antibody-binding mechanism in the Overvelde study. The rationale for using biotin–streptavidin was that its high binding affinity (ka z 1015/M) would maximize any differences between an adherent and non-adherent MBs. In contrast, fluorescein–anti-fluorescein has a much lower binding affinity (ka z 1010/M) which could also account for some of the observed differences. The contribution of binding affinity to MB acoustic behavior has not been explored and will be a focus of a future investigation. The local MB confining conditions are also known to have a dramatic effect on the resulting bubble dynamics, wall stiffness can either increase or decrease the resonance frequency of an MB positioned near it (Doinikov et al. 2011; Martynov et al. 2009; Qin and Ferrara 2007; Sassaroli and Hynynen 2005). Rigid
boundaries have the effect of decreasing the resonance frequency, whereas compliant boundaries have the opposite effect. Furthermore, the relative size of the vessel compared with the MB has also been shown to enhance this change in resonance frequency. In this study, the MBs are confined in a 200-mm cellulose capillary fiber compared with the Overvelde study, which used the Opticell culture dishes (polystyrene membranes with 2 mm of clearance). A number of size-dependent effects become more pronounced because the materials are different, and hence the material properties and effects on MB dynamic characteristics, and the capillary fiber is much closer in size to the MBs. One must also examine what it is that the studies are actually gathering as their data sets. The present study examines the scattered pressure and specifically the second harmonic signal generation as an indicator of resonance. In comparison, the Overvelde study is examining the radial oscillations and specifically the maximum radial excursion. By nature, this is predominately a measure of the fundamental MB response. As a further complication, these data are obtained via a high-speed camera limited to a single orthogonal view. As previously stated, an MB near a boundary can experience significant asymmetry in oscillation. Although it has been demonstrated that it is possible to derive the pressure-time trace from the radius-time curve by assuming a purely radial oscillation (Sijl et al. 2011), it has also been shown that this relationship can break down when examining higher-frequency components above the fundamental response. MB response to repeated insonation Figure 9 indicates that resonance plays an important role in the stability and longevity of MBs. This stands to
Characterization of adherent microbubbles d J. CASEY et al.
reason when one considers that a bubble at resonance will experience the widest fluctuations in size and shape deformations. This finding is supported by Guidi et al. (2010) and Vos et al. (2009), who used a high-speed camera approach to show that phospholipid-shelled MBs undergo acoustically induced deflation after repeated insonation and that this is exacerbated near resonance. The shift in resonance size shown in the second harmonic scattering is also clearly evident in both parts of Figure 9 by the shift in disruption threshold to larger MBs for adherent MBs. Unlike the second harmonic generation, however, there is no distinct disruption peak associated with the resonance frequency. The stability index indicates that bubbles smaller than the resonance radius undergo severe disruption even at the relatively low acoustic pressure (specifically the small radii MBs of Fig. 9); the caveat is that this is true only for these specific experimental conditions. It could be that given larger MBs and a lower driving frequency, the bubbles smaller than the resonance radius might be robust enough to survive. The use of the broadband destructive pulse activated a wider size distribution of MBs as demonstrated by the increase in disruption threshold for both adherent and non-adherent MBs. In the future, however, it might be useful to consider using either a broader band or a transducer transmitting a lower center frequency for the activation of a more diverse population of the MBs. With the current setup, it can be seen that the majority of MBs interrogated are still larger than the size range affected by the ultrasound in both the second harmonic and stability index data. A number of MBs show that there has been an increase in the total scattering to successive pulse (a stability index .1). As already described, MBs are known to exhibit acoustically induced deflation in response to ultrasound (Guidi et al. 2010). By this mechanism, if an MB with a radius initially greater than the resonance radius was insonated, each successive pulse could deflate the MB, driving its radius toward the resonance size. In this case and with reference to Figure 8, it can be seen that for little change in fundamental scattering as the MB approaches resonance, the second harmonic contribution to total scattering would result in an increase in total scattering and hence a value greater than 1 for the stability index. Directions for future investigation The shift in acoustic frequency detected could have potential for the development of novel imaging strategies for the differentiation of adherent MBs. At present, the detected shift in resonance frequency is small and subject to variation between MBs (even of similar size). In most bulk or non-single bubble techniques, these subtle differ-
913
ences will be masked by the response from the polydisperse MB population. Currently there is a significant effort developing methods for the production of monodisperse populations of MBs (Talu 2007; Talu et al. 2006; Xu et al. 2006). With tightly controlled MB physical characteristics (e.g., radius, shell composition), the variability between MBs of similar sizes will be reduced and the MB radius can be tailored to the specific imaging parameters and make maximal use of any resonance frequency shift. Furthermore, the examination of sub-harmonics (SHs) rather than higher harmonics might be worth considering. SHs are produced at half the insonation frequency and have been shown to be at a maximum when an MB is driven at twice its resonance frequency (Katiyar and Sarkar 2010). If the shift in resonance observed here is also evident in the SH response of MBs, this would effectively result in the difference in resonance radius being exaggerated twofold, allowing for easier differentiation. CONCLUSIONS The acoustic setup for detailed interrogation of single MBs under controlled conditions is presented. The adherence of MBs to a capillary fiber wall via a biotin–streptavidin bond can be seen to increase the MB resonance frequency by shifting the radius of peak generation of second harmonic signals. The fundamental scattering properties of the MBs are unchanged by the presence of adhesion. MB stability is heavily dependent on the resonance properties, with resonance bubbles being significantly more susceptible to disruption than MBs situated away from the resonance radius. The change in resonance frequency shown by the second harmonic generation is also highlighted by the bubble disruption. Acknowledgments—The authors would like to acknowledge the contributions from Jonathan Loughran, Dr. Richard Browning and Dr. Emma Behjat. This work was supported by Engineering and Physical Sciences Research Council grant number EP/G038163/1.
REFERENCES Anderson DR, Tsutsui JM, Xie F, Radio SJ, Porter TR. The role of complement in the adherence of microbubbles to dysfunctional arterial endothelium and atherosclerotic plaque. Cardiovasc Res 2007; 73:597–606. Borden MA, Pu G, Runner GJ, Longo ML. Surface phase behavior and microstructure of lipid/PEG-emulsifier monolayer-coated microbubbles. Colloids Surf B Biointerfaces 2004;35:209–223. Caskey CF, Kruse DE, Dayton PA, Kitano TK, Ferrara KW. Microbubble oscillation in tubes with diameters of 12, 25, and 195 microns. Appl Phys Lett 2006;88:033902. Caskey CF, Stieger SM, Qin S, Dayton PA, Ferrara KW. Direct observations of ultrasound microbubble contrast agent interaction with the microvessel wall. J Acoust Soc Am 2007;122:1191–1200. Chomas JE, Dayton P, May D, Ferrara K. Threshold of fragmentation for ultrasonic contrast agents. J Biomed Opt 2001;6:141–150.
914
Ultrasound in Medicine and Biology
Couture O, Bannouf S, Montaldo G, Aubry J, Fink M, Tanter M. Ultrafast imaging of ultrasound contrast agents. Ultrasound Med Biol 2009;35:1908–1916. Dayton JJR. Molecular ultrasound imaging using microbubble contrast agents. Front Biosci 2007;12:5124–5142. Doinikov A, Leila A, Ayache B. Acoustic scattering from a contrast agent microbubble near an elastic wall of finite thickness. Phys Med Biol 2011;56:6951. Doinikov AA, Haac JF, Dayton PA. Resonance frequencies of lipidshelled microbubbles in the regime of nonlinear oscillations. Ultrasonics 2009;49:263–268. Dollet B, van der Meer SM, Garbin V, de Jong N, Lohse D, Versluis M. Nonspherical oscillations of ultrasound contrast agent microbubbles. Ultrasound Med Biol 2008;34:1465–1473. Frinking PJA, Gaud E, Brochot J, Arditi M. Subharmonic scattering of phospholipid-shell microbubbles at low acoustic pressure amplitudes. IEEE Trans Ultrason Ferroelectr Freq Control 2010;57: 1762–1771. Garbin V, Cojoc D, Ferrari E, Di Fabrizio E, Overvelde MLJ, van der Meer SM, de Jong N, Lohse D, Versluis M. Changes in microbubble dynamics near a boundary revealed by combined optical micromanipulation and high-speed imaging. Appl Phys Lett 2007; 90:114103. 3. Guidi F, Tortoli P, Vos HJ, de Jong N. Simultaneous optical and acoustical observation of microbubbles behaviour. IEEE Ultrasonics Symposium, Vancouver, Canada, 2006. Guidi F, Vos HJ, Mori R, De Jong N, Tortoli P. Microbubble characterization through acoustically induced deflation. IEEE Trans Ultrason Ferroelectr Freq Control 2010;57:193–202. Helfield BL, Cherin E, Foster FS, Goertz DE. Investigating the subharmonic response of individual phospholipid encapsulated microbubbles at high frequencies: A comparative study of five agents. Ultrasound Med Biol 2012;38:846–863. Katiyar A, Sarkar K. Excitation thresholds for subharmonic response of ultrasound contrast microbubbles. J Acoust Soc Am 2011;130: 3137–3147. Lindner JR. Assessment of inflammation with contrast ultrasound. Prog Cardiovasc Dis 2001;44:111–120. Lindner JR. Microbubbles in medical imaging: current applications and future directions. Nat Rev Drug Discov 2004;3:527–533. Lindner JR, Dayton PA, Coggins MP, Ley K, Song J, Ferrara K, Kaul S. Noninvasive imaging of inflammation by ultrasound detection of phagocytosed microbubbles. Circulation 2000;102:531–538. Marmottant P, van der Meer S, Emmer M, Versluis M, de Jong N, Hilgenfeldt S, Lohse D. A model for large amplitude oscillations of coated bubbles accounting for buckling and rupture. J Acoust Soc Am 2005;118:3499–3505. Martynov S, Stride E, Saffari N. The natural frequencies of microbubble oscillation in elastic vessels. J Acoust Soc Am 2009;126:2963–2972. Morgan KE, Allen JS, Dayton PA, Chomas JE, Klibaov AL, Ferrara KW. Experimental and theoretical evaluation of microbubble behavior: effect of transmitted phase and bubble size. IEEE Trans Ultrason Ferroelectr Freq Control 2000;47:1494–1509. Mulvana H, Stride E, Tang M-X, Hajnal JV, Eckersley RJ. The influence of gas saturation on microbubble stability. Ultrasound Med Biol 2012;38:1097–1100. Needles A, Couture O, Foster FS. A method for differentiating targeted microbubbles in real time using subharmonic micro-ultrasound and interframe filtering. Ultrasound Med Biol 2009;35:1564–1573. Ohl SW, Klaseboer E, Khoo BC. The dynamics of a non-equilibrium bubble near bio-materials. Phys Med Biol 2009;54:6313–6336. Overvelde M, Garbin V, Dollet B, de Jong N, Lohse D, Versluis M. Dynamics of coated microbubbles adherent to a Wall. Ultrasound Med Biol 2011;37:1500–1508. Overvelde M, Garbin V, Sijl J, Dollet B, de Jong N, Lohse D, Versluis M. Nonlinear shell behavior of phospholipid-coated microbubbles. Ultrasound Med Biol 2010;36:2080–2092. Overvelde M. PhD Thesis. Ultrasound Contrast Agents Dynamics of Coated Bubbles, University of Twente, Netherlands, 2010.
Volume 39, Number 5, 2013 Patil AV. A 3D nonlinear FEA model for simulation of microbubble behavior. IEEE International Ultrasonics Symposium Proceedings, Beijing, China, 2008. Pitt WG. Ultrasonic drug delivery—A general review. Expert opin Drug Deliv 2004;1:37–56. Qin S, Ferrara KW. The natural frequency of nonlinear oscillation of ultrasound contrast agents in microvessels. Ultrasound Med Biol 2007;33:1140–1148. Sassaroli E, Hynynen K. Resonance frequency of microbubbles in small blood vessels: A numerical study. Phys Med Biol 2005;50: 5293–5305. Sboros V, Glynos E, Pye SD, Moran CM, Butler M, Ross J, Short R, McDicken WN, Koutsos V. Nanointerrogation of ultrasonic contrast agent microbubbles using atomic force microscopy. Ultrasound Med Biol 2006;32:579–585. Sboros V, Moran CM, Pye SD, McDicken WN. The behaviour of individual contrast agent microbubbles. Ultrasound Med Biol 2003;29: 687–694. Sboros V, Pye SD, MacDonald CA, Gomatam J, Moran CM, McDicken WN. Absolute measurement of ultrasonic backscatter from single microbubbles. Ultrasound Med Biol 2005;31:1063–1072. Sennoga CA, Mahue V, Loughran J, Casey J, Seddon JM, Tang M, Eckersley RJ. On sizing and counting of microbubbles using optical microscopy. Ultrasound Med Biol 2010;36:2093–2096. Sijl J, Vos HJ, Rozendal T, de Jong N, Lohse D, Versluis M. Combined optical and acoustical detection of single microbubble dynamics. J Acoust Soc Am 2011;130:3271–3281. Talu E. Tailoring the size distribution of ultrasound contrast agents: Possible method for improving sensitivity in molecular imaging. Mol Imaging 2007;6:384–392. Talu E, Hettiarachchi K, Nguyen H, Lee AP, Powell RL, Longo ML, Dayton PA. Lipid-stabilized monodisperse microbubbles produced by flow focusing for use as ultrasound contrast agents. Proceedings of the IEEE Ultrasonics Symposium, Vancouver, Canada, 2006. Tu J, Guan J, Qiu Y, Matula TJ. Estimating the shell parameters of SonoVue microbubbles using light scattering. J Acoust Soc Am 2009; 126:2954–2962. Tu J, Swalwell JE, Giraud D, Cui W, Chen W, Matula TJ. Microbubble sizing and shell characterization using flow cytometry. IEEE Trans Ultrason Ferroelectr Freq Control 2011;58:955–963. van der Meer SM, Versluis M, Lohse D, Chin CT, Bouakaz A, de Jong N. The resonance frequency of SonoVue; as observed by high-speed optical imaging. Ultrasonics Symposium, 2004 IEEE, Montreal, Canada, 2004. pp. 343–345 Vol. 1. Vos HJ, de Jong N, Mori R, Viti J, Guidi F, Tortoli P, Sijl J. Ultrafast framing optical and acoustical recording of nonlinearly oscillating microbubbles. Ultrasonics Symposium (IUS), 2009 IEEE International, Rome, Italy, 2009. pp. 1817–1820. Vos HJ, Dollet B, Bosch JG, Versluis M, de Jong N. Nonspherical vibrations of microbubbles in contact with a wall–a pilot study at low mechanical index. Ultrasound Med Biology 2008;34:685–688. Wijkstra H, Smeenge M, Rosette J, Pochon S, Tardy-Cantalupi I, Tranquart F. Targeted microbubble prostate cancer imaging with BR55. 17th European Symposium on Ultrasound Contrast Imaging, Rotterdam, Netherlands, 2012. Willmann JK, Paulmurugan R, Chen K, Gheysens O, Rodriguez-Porcel M, Lutz AM, Chen IY, Chen X, Gambhir SS. US imaging of tumor angiogenesis with microbubbles targeted to vascular endothelial growth factor receptor type 2 in Mice1. Radiology 2008;246:508–518. Xie F, Tsutsui JM, Lof J, Unger EC, Johanning J, Culp WC, Matsunaga T, Porter TR. Effectiveness of lipid microbubbles and ultrasound in declotting thrombosis. Ultrasound Med Biol 2005; 31:979–985. Xu JH, Li SW, Chen GG, Luo GS. Formation of monodisperse microbubbles in a microfluidic device. AIChE J 2006;52:2254–2259. Zhao S. Detection of echoes from adherent targeted microbubbles. IEEE Ultrasonics Symposium, Vancouver, Canada, 2006. Zhao S, Kruse DE, Ferrara KW, Dayton PA. Selective imaging of adherent targeted ultrasound contrast agents. Phys Med Biol 2007; 52:2055–2072.
http://dx.doi.org/10.1016/j.ultrasmedbio.2012.12.007
d
Original Contribution SINGLE BUBBLE ACOUSTIC CHARACTERIZATION AND STABILITY MEASUREMENT OF ADHERENT MICROBUBBLES JONATHAN CASEY,* CHARLES SENNOGA,* HELEN MULVANA,* JO V. HAJNAL,* MENG-XING TANG,y and ROBERT J. ECKERSLEY* * Imaging Sciences Department, Imperial College, London, UK; and y Department of Bioengineering, Imperial College, London, UK (Received 31 January 2012; revised 10 December 2012; in final form 11 December 2012)
Abstract—This article examines how the acoustic and stability characteristics of single lipid-shelled microbubbles (MBs) change as a result of adherence to a target surface. For individual adherent and non-adherent MBs, the backscattered echo from a narrowband 2-MHz, 90-kPa peak negative pressure interrogation pulse was obtained. These measurements were made in conjunction with an increasing amplitude broadband disruption pulse. It was found that, for the given driving frequency, adherence had little effect on the fundamental response of an MB. Examination of the second harmonic response indicated an increase of the resonance frequency for an adherent MB: resonance radius increasing of 0.3 ± 0.1 mm for an adherent MB. MB stability was seen to be closely related to MB resonance and gave further evidence of a change in the resonance frequency due to adherence. (E-mail: [email protected]) Ó 2013 World Federation for Ultrasound in Medicine & Biology. Key Words: Ultrasound, Microbubbles, Targeting, Adherence, Single bubbles, Resonance, Stability, Molecular imaging.
2004). By this attachment, the MB is brought to within tens of nanometers of the vasculature wall, which has been shown to affect the dynamic response of an MB to ultrasound (Caskey et al. 2006, 2007; Dollet et al. 2008; Garbin et al. 2007; Martynov et al. 2009; Ohl et al. 2009; Overvelde 2010; Sassaroli and Hynynen 2005). One of the main challenges associated with the use of targeted MBs for molecular imaging is how the signals from the attached MBs are differentiated from those of the free flowing or unattached MBs. The first approach typically used is that of clearance imaging. Unbound MBs are typically cleared from the circulatory system in approximately 10 min leaving the bound MBs for imaging (Lindner 2001). The main disadvantage with this technique is that although some MBs are retained at the target site, this number too decreases throughout the clearance period. Ideally, imaging should be performed at the time point of maximal MB retention and has directed many approaches to focus on image processing techniques to achieve this differentiation. Because of their compressible nature, MBs generate substantially more scattering than tissue and in particular generate more harmonic signal, either higher or subharmonics, even at low acoustic pressures (Frinking et al. 2010; Guidi et al. 2010). Such features allow MBs
INTRODUCTION Targeted microbubbles (MBs) have shown potential in allowing the use of ultrasound in a variety of new applications (Dayton 2007; Lindner 2004). By functionalizing the shell of the MBs to target specific markers expressed in the endothelium, the contrast agent can be used to highlight areas of interest and have been used in vivo to image inflammation (Lindner et al. 2000), angiogenesis (Willmann et al. 2008), atherosclerosis (Anderson et al. 2007) and thrombosis (Xie et al. 2005). Currently, targeted contrast agents for the molecular imaging of prostate cancer are undergoing initial human trials (Wijkstra et al. 2012). Their use as targeted drug delivery vehicles is also under investigation (Pitt 2004). These targeting ligands—be they antibodies, proteins or peptides dependent on the pathology to be targeted—are situated on the surface architecture of the MB, either on the shell itself or more commonly at the end of a spacer, typically a poly-ethylene glycol (PEG) spacer (Borden et al.
Address correspondence to: Robert J. Eckersley, Biomedical Engineering Department, Division of Imaging Sciences, King’s College London, St. Thomas’ Hospital, London, United Kingdom. E-mail: [email protected] 903
904
Ultrasound in Medicine and Biology
to be distinguished from tissue by image subtraction (Zhao et al. 2007) or examination of the harmonic signals generated. The differentiation between attached and unattached bubbles is then conducted by temporal filtering. An attached bubble will appear as a stationary scatterer, whereas an unattached bubble will appear as a moving scatterer; therefore, stationary bubbles can be detected by applying a low-pass filter to remove bubbles in motion. Several authors have successfully demonstrated variations of this technique in vitro (Needles et al. 2009; Zhao 2006; Zhao et al. 2007); however, in vivo a number of additional complications, such as body or tissue movement, make these techniques more difficult. A more efficient and selective imaging strategy could be developed if the signals generated by attached MBs showed unique acoustic characteristics that could then be identified specifically. The acoustic characteristics of MBs under a variety of conditions must be determined precisely so that specific imaging protocols can be devised. It is known that the local conditions can have a significant effect on MB response. Increasing values of wall rigidity decrease MB resonance frequency. When an elastic wall is modeled in these cases, it can have the effect of increasing or decreasing the resonance frequency depending on the specific material properties of both the wall and the surrounding medium. In most physiologically analogous situations, this has been shown to increase MB resonance frequency (Caskey et al. 2007; Doinikov et al. 2011; Martynov et al. 2009; Qin and Ferrara 2007). Patil et al. (2008) conducted a 3-D finite element examination of an MB adherent to a rigid wall in comparison with a free, unconstrained MB. In addition to showing many of the shape fluctuations as seen in highspeed imaging experiments (Dollet et al. 2008; Vos et al. 2008), the model also predicted a shift in resonance frequency as a result of adherence, having the effect of lowering the resonance frequency and suppressing some of the second harmonic signal generation. These observations have been supplemented by a high-speed imaging study (Overvelde et al. 2011); however, acoustic verification of these results has not yet been presented. Furthermore, MB destruction thresholds and unstable mode oscillations have been shown to be affected by the presence and characteristics of nearby boundaries; however, the presence and effects of binding on MB stability are not fully understood. Couture et al. (2009) demonstrated that there is also a difference in the dissolution time of adherent MBs compared with non-adherent MBs after ultrasound exposure. Adherent MBs were shown to dissolve at a significantly faster rate than their nonadherent counterparts. This finding could provide a differentiation technique based on the disruption of MBs; however, this would require high temporal resolution to be achieved.
Volume 39, Number 5, 2013
Any difference, such as changes to resonance frequency, detected between adherent and non-adherent MB acoustic signatures is expected to be subtle and size dependent. As such, bulk acoustic measurements—for which scattering and attenuation measurements are performed on bulk suspensions of MBs—are unlikely to elucidate the important details. Consequently, the acoustics characteristics of single MBs in varying conditions will be investigated in this work. The acoustic characterization of individual MBs has been implemented by a number of authors (Guidi et al. 2010; Sboros et al. 2003, 2005; Guidi et al. 2006; Vos et al. 2009) who have highlighted phenomena including acoustically induced deflation, compression and expansion-only behavior and changes in bubble stability when confined in small vessels. The objective of this study is to perform an acoustic investigation of the effect that binding an MB to a surface has in comparison with an unbound MB in otherwise similar conditions (i.e., near a boundary yet unattached). Specific attention will be given to both the resonance characteristics of the MBs and changes in bubble response to repeated insonation. MATERIALS AND METHODS Experimental setup The equipment (Fig. 1) consisted of a pair of matched transducers (V380, Panametrics, Olympus Industrial, Southend-on-Sea, UK) (4 MHz center frequency, 26 dB bandwidth 86%, focal length of 75 mm) focused to a central point in conjunction with a 403 water-immersible objective (LUMPlanFL N40xW, Olympus Medical, Southend-on-Sea, UK). Positioned at the focus was a capillary fibre (200 mm diameter; RC55 8/200, Membrana, Wuppertal, Germany) through which the MB suspension was flowed. Sham experiments, conducted without the presence of MBs, showed that the fiber did not scatter sound at a detectable level using the current setup and was therefore considered both acoustically and optically transparent. The fiber was mounted on a three-axis translational stage (562-XYZ ULTRAlignÔ Precision Linear Stage, Newport Corporation, Irvine, CA, USA) for accurate positioning. Note that to minimize acoustic interference, particular care must be taken over the alignment of the transducers and optics. The transducers were positioned approximately 10 degrees off axis from the objective. Similarly, the angles between the two transducers were unequal (positioned approximately 20 and 45 degrees from the normal for receiving and transmitting transducers, respectively) to minimize direct reflections. An additional complication was that, because of the required proximity of the objective to the capillary fiber (limited by the objective working distance of 3.3 mm), the presence of a scatterer in the fiber caused complex interference in the received signals. To remedy this, once an optical image had been
Characterization of adherent microbubbles d J. CASEY et al.
905
Fig. 1. Schematic of the experimental setup.
acquired to size the MB, the objective was removed from the acoustic field before taking the acoustic measurements. The transmitted signal was generated in an arbitrary waveform generator (Sony Tektronix AWG2021) and amplified through a power amplifier (2100L; E&I, Rochester, NY, USA). The scattered signal was amplified by a pulser/receiver operated in receive mode (Panametrics-NDT 5800) and displayed via a digital oscilloscope (Sony Tektronix TDS7154, Tektronix UK Ltd., Bracknell, UK). The image from the objective was directed to a digital camera (Canon ProShot G5) through a 45-degree mirror and focusing lens for subsequent sizing. Optical calibration was performed using latex sizing beads (3, 5 and 10 mm; Beckman Coulter, Brea, CA, USA). The beads were optically imaged and sized based on image intensity to determine their radius. To determine the robustness of the sizing technique, multiple images of populations of the three different beads were examined. This examination resulted in an average error of 0.08 mm and a maximum error of 0.14 mm independent of particle size. This approach is based on the work of Helfield et al. (2012). Final optical calibration was found to be 20 pixels/mm. The entire system was controlled via a desktop computer and an internally developed MATLAB program (Mathworks, Cambridge, UK). The gas saturation of the solution has been shown to affect the MB stability and the reproducibility of measurements (Mulvana et al. 2012). In accordance with these findings, filtered, gas-saturated water was used for all experiments. The water was filtered via reverse osmosis (euRO 20; Triple Red Ltd., Long Crendon, Buckinghamshire, UK) and was considered deionized. All experiments were performed at 20 C. Microbubble preparation Microbubbles were prepared by sonication (Misonix Sonicator 3000; settings: 21 kHz, 165 W, 30 s) of
an octo-fluoropropane–saturated aqueous suspension of distearoyl-phosphatidylcholine, distearoyl-phosphatidylethanolamine-PEG2000-biotin (DSPE-PEG2000-Biotin) and poly(ethylene glycol)-monostearate (PEG40-stearate). Non-targeted MBs (control) were prepared similarly, substituting DSPE-PEG2000-Biotin with methoxy-poly (ethylene glycol)2000 distearoyl-phosphatidylethanolamine (DSPE-mPEG2000). MB size distribution and concentration were measured to have a mean size diameter of 2.4 6 0.4 mm and a number concentration of 1.2 3 109 MBs/mL using optical microscopy (Sennoga et al. 2010). To conduct these experiments, it was vital to ensure that only a single bubble was present at the acoustic focus. To achieve this requirement, the MB solution was diluted by approximately 1:100,000 with some variation introduced because of bubble buoyancy, sampling errors and variations in the initial concentration of the MB solution. The transducer focal region was measured to be approximately 2.4 mm in diameter. The capillary fiber was checked optically to a distance of .1.5 mm either side of the target bubble to verify that no other bubbles were present. Upon injection, MBs rise to the upper inner surface of the capillary fibre, which as stated in the introduction brings with it some profound changes in bubble response. Because this change will affect both the adherent and nonadherent MBs, it was deemed acceptable and favorable, allowing direct comparison of bound and unbound bubble without any changes in local conditions or geometry.
Tube preparation Capillary fibers were cleaned with sterile phosphatebuffered saline (Sigma-Aldrich, Dorset, UK) and flooded with unlabeled streptavidin (Invitrogen Life
906
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
Fig. 2. Pulse definition; 2-MHz, 10-cycle narrowband interrogation pulse (top left); broadband 3.5-MHz, two-cycle destruction pulse (top right). Whole-pulse sequence showing the first five interrogation pulses [bottom (a)], increasing ramp of destruction pulses interleaved with the interrogation pulses [bottom (b)].
Technologies, Paisley, UK) solution at a concentration of 0.25 mg/mL before incubation overnight at 4 C. Acoustic pulse characteristics Accurate and detailed transducer characterization was essential to the success of this investigation. To obtain the frequency response spectrum, a plane reflector was positioned at the focus of the receive transducer. Operating the transducer via the pulser/receiver, the transducer was driven with an impulse and the subsequent response was recorded (details of bandwidth and center frequency given in the Experimental Setup section). This response is the frequency response after passing through both the transducers twice (transmitting and receiving). Assuming transducer reciprocity, the response for travel in one direction was taken as the square root of the full response. Spatial and pressure characterization of the ultrasound field was conducted using a calibrated 1-
mm–diameter needle hydrophone (HPM1/1; Precision Acoustics, Dorset UK). Each MB was insonated with a pulse sequence split into two distinct phases. Phase one (Fig. 2a) consisted of five, 2-MHz, 10-cycle pulses at a peak negative pressure (PNP) at the focus of 90 kPa. Phase two (Fig. 2b) consisted of alternating two-cycle, 3.5-MHz, center-frequency broadband pulses and the same 2-MHz narrowband pulse as in phase one for 10 repetitions (20 pulses in total). After each repetition, the broadband pulse amplitude was increased in increments of approximately 45 kPa from 50–450 kPa PNP. All pulses are sent at a pulse repetition frequency of 100 Hz. The rationale for this pulse sequence was twofold. The first narrowband pulse will give the bubble’s acoustic behavior at the bubble size imaged by optical microscope. The repeated pulses will then show how the bubble responds to repeated exposure. It is
Characterization of adherent microbubbles d J. CASEY et al.
907
Fig. 3. Typical acoustic responses for MBs at resonance (left) and away from resonance (right). Top panes show sizing photos, and the middle and bottom panes show the bubble response in the time and frequency domains, respectively. Scale bar 5 5 mm. MBs 5 microbubbles.
understood that MB stability is significantly affected by resonance (Chomas et al. 2001); it was intended that the first five pulses should provide detailed information on the acoustic and stability characteristics of MBs at or near resonance for a given narrowband excitation. Because the size distribution of bubbles examined is dependent on their initial size distribution and longevity after extreme dilution, the technique is biased toward larger bubbles. For bubbles with radii away from that of the resonance size, it has been shown that little disruption occurs (Guidi et al. 2010). Therefore, in an effort to maximize the data obtainable from MBs away from resonance, broadband pulses of increasing amplitude were used.
Data analysis The data for each bubble was split into single pulses, corrected for amplifier gain, zero offset and windowed using a hamming window. The data were then corrected for the frequency response characteristics of the transducer.
Figure 3 shows both the time and frequency domain responses recorded for two adherent MBs of different sizes. The left panels show the response of an MB near resonance. Note that for a 2-MHz insonation pulse, the resonance radius for a lipid-encapsulated bubble is predicted to be 1.8 mm (Morgan et al. 2000, van der Meer et al. 2004). This near resonance response is clearly demonstrated by the strong second harmonic component in the frequency spectrum. The right panels show the response of a bubble that is greater than the resonance size for a 2-MHz insonation pulse. Fundamental scattering can be seen to be increased, and the generation of the second harmonic signal is diminished. In the results that follow, scattering power is defined as the total area under the frequency spectrum curve normalized by the noise floor (see eqn [1]). Similarly, the scattering power of the separate harmonics was taken as the band-pass area under the harmonic peak (0.5 MHz either side of the peak frequency). By tracking the changes in scattering power of the various harmonics as a function of bubble size, a detailed picture of MB acoustic characteristics can be determined.
908
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
"Pfhigh Scattering PowerðdBÞ 5 20 log10
where fhigh and flow refer to the upper and lower bounds of the frequency range, corresponding to the peak frequency (60.5 MHz). To evaluate the change in bubble response to repeated insonation, the total scattering power was tracked as a function of interrogation pulse number. Figure 4 shows the normalized scattered power against the pulse number for three bubbles undergoing different changes in response. Scattered power is normalized against the response from the first pulse. Three distinct modes of MB response are visible. The green trace shows an MB that does not alter in response power over the entire pulse sequence as demonstrated by a normalized power remaining at 1. The blue trace shows an MB that undergoes a reduction in scattering response immediately after the first insonation pulse. The red trace shows an MB that does not exhibit a change until after the first five interrogation pulses. In reference to Figure 2, this indicates that the increasing ramp of broadband disruption pulses is responsible for the change in response. To parameterize this change in response, the area under each response curve was found and normalized against a constant response of 1 (i.e., the response of an MB exhibiting no change); this gives a single value per MB. A value of 1 is equivalent to an MB with no change in response. A value of less than 1 shows a decrease in response. When plotted against initial MB radius, the
flow signal amplitude spectrumðf Þ Pfhigh flow noise amplitude spectrumðf Þ
# (1)
size-dependent change in response was obtained. This parameter will henceforth be referred to as the ‘‘stability index.’’ Individual data points were subject to a large degree of variation, the cause of which will be discussed later. To extract trends within the data, the mean and standard deviation of a five-point moving window was calculated for each data point in both the sizing (x-axis) and parameter (y-axis) directions. Note that the end data points were subject to smaller widow sizes; the first and last data points used a three-point window, and the second and penultimate points used a four-point window. On subsequent figures using this windowing, mean data are displayed as a line with shaded regions representing the standard deviation of the mean. Raw data are displayed as discrete points. To show the statistical significance between the comparison trends, the data are binned into 0.2-mm bins (conservatively above the maximum sizing error). This process results in bins ranging in size up to 12 data sets per bin depending on the spread of MB sizes interrogated. The data are then tested using a two-tailed unpaired t-test to determine whether the two sets of data are significantly different. The significance level is set to p . 0.05, and the regions of significant difference are highlighted in the relevant figures. RESULTS
Fig. 4. Normalized total scattering as a function of interrogation pulse number to show bubble response as a function of pulse repetition. The green line shows a bubble with no change in response over the pulse sequence. The blue line shows a bubble exhibiting a change in response subsequent to the first insonation. The red line shows a bubble exhibiting a change in response as the destructive broadband pulses are applied (post pulse 5).
Scattered acoustic power The following data present the results from all four experimental conditions tested: (i) untargeted MBs, uncoated capillary fiber (UTUC); (ii) untargeted MBs, coated capillary fiber (UTC); (iii) targeted MBs, uncoated fiber (TUC); and (iv) targeted MBs, coated fiber (TC). The number of data points for each of the test cases were: TC, n 5 41; UTUC, n 5 44; TUC, n 5 49; and UTC, n 5 45. By comparison of these cases, the effect that the functionalization of both the bubbles and the capillary fibre has on MB dynamic response can be elucidated. For display purposes, each case is shown individually, and for comparative analysis the mean trend lines are superimposed on each other. Figure 5 shows the fundamental scattering results. In general, good agreement can be observed between all testing regimes, especially through the resonance region where the greatest variation was anticipated (p . 0.05
Characterization of adherent microbubbles d J. CASEY et al.
909
Fig. 5. Fundamental scattering as a function of bubble radius for the four testing regimens, showing the three nonadherent or control cases of untargeted–uncoated (top left), targeted–uncoated (top right), untargeted–coated (middle left) and the adherent or targeted–coated (middle right) regimen. Mean trend lines are superimposed to show the comparison (bottom); p . 0.05 for the entire radius range.
for the entire range radius). Figure 6 displays the second harmonic scattering data. Good agreement is observed among the three non-adherent cases (i.e., UTUC, TUC and UTC; p . 0.05) both in terms of resonance radii and scattering power. The slightly lower peak values for the UTUC case could be explained by the limited number of data points within this region (Fig. 5, top left). The adherent MBs are statistically different (p , 0.05) from all three non-adherent cases in the radius range of 1.8–2.0 mm. To establish the radii associated with peak second harmonic generation, the data were fitted with a polynomial curve. Figure 7 shows this curve fitting for the adherent MBs. The radius at which maximum second harmonic generation occurs is indicated by the solid line (R 5 1.86 mm). To provide confidence bounds for this peak, the mean residual was subtracted from the peak second harmonic scattering power; this provided a range of MB radii of 1.75–1.97 mm. These data have been collated for all testing regimens in Table 1. Given the similarities among the three unbound cases, the following sections consist of comparisons of the adherent (TC) and the non-adherent (TUC) MBs to identify differences among otherwise identical bubbles with or without target receptors. Figure 8 shows the MB scattering power as a function of MB radius for fundamental scattering (top) and second harmonic scattering (bottom). The fundamental scattering is minimal below the resonance radius. Through the resonance radius for a 2-MHz insonation pulse (1.8 mm based on current theory for phospholipid shelled MBs; Morgan et al. 2000; van der Meer et al. 2004), there is an increase in the fundamental scattering until a radius of 2 mm, at which point the rate of increase
levels out. This trend is consistent in both the adherent and non-adherent MBs. There is no statistical difference between the trends. The second harmonic scattering power shows a maximum at approximately where one would expect the resonance radius to occur. Peak maxima were extracted (Table 1) and showed an adherent MB resonance radius of 1.86 mm and a non-adherent resonance radius of 1.60 mm; the two data sets have been shown to be statistically different in the range of 1.8–2.0 mm (p , 0.05). Away from resonance effects (.2.5 mm), the response characteristics between the two bubble types are in agreement with diminished second harmonic generation. MB response to repeated insonation Figure 9 (top) shows the bubble stability index after the first five insonation pulses. A stability index of 1 refers to a bubble whose response is invariant with repeated insonation. A stability index less than 1 indicates that the MB response decreases with subsequent insonation pulses. A number of bubbles can also be seen to have a stability index of greater than 1, indicating an increase in acoustic response to successive insonations; the possible reasons for this are explained in the Discussion. MBs near the peak second harmonic generation as seen in Figure 8 for both adherent and non-adherent MBs are subject to the most disruption. Upon exposure to the first five interrogation pulses, non-adherent MBs first experience a decrease in response at radii ,1.8 mm; this is in comparison with adherent MBs that experience this response at a larger initial radius (,2.2 mm; p , 0.05 in the range of 1.8–2.0 mm). The size below which the MB response is seen to decrease will be called
910
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
Fig. 6. Second harmonic scattering as a function of bubble radius for the four testing regimens showing the three nonadherent or control cases of untargeted–uncoated (top left), targeted–uncoated (top right), Untargeted–coated (middle left) and the adherent or targeted–coated (middle right) regimen. Mean trend lines are superimposed to show the comparison (bottom). All three non-adherent regimes have a p . 0.05 throughout the radius range examined. The adherent case differs (p , 0.05) from the non-adherent cases in the radius range of 1.8–2.0 mm.
the ‘‘disruption threshold.’’ As with the second harmonic generation, for MBs with a radius greater than 2.5 mm, the MBs can be seen to be essentially invariant with pulse repetition for both adherent and non-adherent MBs (p . 0.05). Figure 9 (bottom) shows the stability index after the bubble has been exposed to the whole pulse sequence (i.e., insonated with the interrogation pulses and the increasing pressure ramp of broadband insonation); in general, it shows the same trend as Figure 9 (top), showing the size dependence on the stability index. Note that the MB disruption threshold for both MB cases has increased, from approximately 1.8–2.0 mm for nonadherent MBs and from 2.2–2.4 mm for adherent MBs. Furthermore, the levels of disruption experienced in
both cases are dramatically increased, and it follows that the curves are significantly different over a wider range (p , 0.05 in the range 1.8–2.4 mm). DISCUSSION Experimental errors and variability The data presented shows a large degree of variability, even for similarly sized bubbles, the cause of which may be attributed to several factors. Although bubble response is strongly dictated by initial bubble radius, the bubble’s structural properties have also been shown to have a significant effects, specifically initial surface tension resulting in differences in both amplitude and shape of oscillation (e.g., compression-expansion– only behavior, bubble destruction and fragmentation; Marmottant et al. 2005; Overvelde et al. 2010). During insonation, the MBs are all exposed to a primary radiation force resulting in bubble motion. Given a sufficiently large motion, this could move the MBs out of the focal region of the transducers, giving rise to misleading results. To certify that this would not be a problem, a number of bubbles were insonated with the full pulse as described earlier, but with the microscope Table 1. Resonance radii as derived from curve fitting to second harmonic generation
Fig. 7. Example of curve fitting to the second harmonic scattering data of adherent MBs for resonance detection. Curve maximum occurs at R 5 1.86 mm, indicated by the solid black line. Dashed lines show the confidence interval for peak detection. MBs 5 microbubbles.
Testing regime
Resonance radius (mm)
Confidence interval (6 mm)
Untargeted–uncoated Targeted–uncoated Untargeted–coated Targeted–coated
1.59 1.60 1.60 1.86
0.11 0.09 0.10 0.11
Characterization of adherent microbubbles d J. CASEY et al.
911
Fig. 8. Scattering power as a function of MB radius for fundamental scattering (top) and second harmonic scattering (bottom). Comparison between adherent MBs (dashed blue line) and non-adherent (targeted MBs in an uncoated capillary fibre, solid red line). MBs 5 microbubbles.
objective still in place to directly observe bubble motion. Maximal bubble motion was observed to be of the order of less than 10 mm. Another possible source of error comes from the presence of nearby bubbles in the acoustic field. Although this problem is minimized by the optical verification that the capillary fiber is clear of either side of the target bubble, only the upper surface of the fiber is scanned for time saving purposes and because any MBs should rise through buoyancy; therefore, it does not mitigate against bubbles being entrained elsewhere on or in the fiber, either trapped by particulate matter or forming because the whole system is immersed in gas saturated water. Typically, any large bubbles of this kind are detectable from the magnitude and shape of the response and discounted; however, for smaller bubbles these responses could be overlooked easily, leading to skewed data. Scattered acoustic response All the bubbles examined displayed similar fundamental response to the 2-MHz insonation pulse, whether they were adherent or non-adherent, indicating that under these conditions the presence of binding does not change the fundamental response of an MB. However, examination of the second harmonic scattering exhibits a difference. The shift in resonance radius between the two cases (1.86 mm for the adherent MB compared with 1.60 mm for the non-adherent MB) indicates an increase in resonance frequency for the adherent MBs. The difference between the resonance radii of the adherent (TC) compared with the three non-adherent MB cases indicates that adhesion causes the shift in resonance radius (and therefore frequency) and is not simply an artifact
of the functionalization of either the capillary fiber or the MB. Although the contribution of the second harmonic can be seen to have a dramatic effect on the level of total scattering, especially around resonance, it does not provide a sufficient difference to delineate the adherent from the non-adherent MBs on its own. This finding in itself is important when trying to implement differentiation protocols in more complex scenarios, implying that detailed frequency content is required rather than just total backscatter in the differentiation of adherent and free MBs. After comparing these results to the findings of Patil et al. (2008) and Overvelde et al. (2011), a number of differences present themselves. All three studies (this study included) show differences in the acoustic characteristics for adherent MBs; however, the way this difference is demonstrated is different in each case. The Overvelde study would appear to be most comparable with the present study, because of similar frequency ranges, bubble type and applied pressures. By insonating both adherent and non-adherent MBs of similar size (radius z 2.1 mm) through a range of frequencies and recording their oscillations via a high speed camera, the frequency of maximum response was found. This frequency of maximal response was then nondimensionalized by normalizing against the resonance frequency of an uncoated free MB of the same radius. The uncoated free resonance frequency given by: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 1 1 2s 2s w w f0uncoated 5 3g P0 1 2 (2) R0 R0 2p rR20 where r is the liquid density (998 kg/m3), R0 is the initial MB radius, g is the polytropic gas exponent (1.07), P0 is
912
Ultrasound in Medicine and Biology
Volume 39, Number 5, 2013
Fig. 9. Stability index for the first five pulses of pulse sequence (top) showing a comparison between adherent MBs (dashed blue line) and non-adherent MBs (solid red line); p , 0.05 in the radius range of 1.8–2.0 mm. Stability index after exposure to the whole pulse sequence (bottom) is shown in a comparison between adherent MBs (dashed blue line) and non-adherent MBs (solid red line); p , 0.05 in the radius range of 1.8–2.4 mm. MBs 5 microbubbles.
the ambient pressure (101 kPa) and sw is the interfacial tension coefficient (0.072 N/m; Doinikov et al. 2009). Given the large discrepancies between the findings of this study and the Overvelde study for seemingly similar acoustic conditions, one must then assume that there is a difference in other experimental conditions or in the way results are analyzed that causes this discrepancy. There are a number of factors that could influence the MB response in these studies. For example, any differences in MB shell chemistry effectively change the MB visco-elastic properties between the two studies. This has the effect of modifying the shape and position of the resonance frequency curve. In addition, it has been postulated that both shell elasticity and viscosity have a size-dependent nature (Sboros et al. 2006; Tu et al. 2009, 2011), further skewing the resonance frequency curves. As an aside to shell chemistry, the binding mechanisms used in the two studies are also different. A biotin–streptavidin ligand pair in this study compared with a fluorescein–anti-fluorescein antibody-binding mechanism in the Overvelde study. The rationale for using biotin–streptavidin was that its high binding affinity (ka z 1015/M) would maximize any differences between an adherent and non-adherent MBs. In contrast, fluorescein–anti-fluorescein has a much lower binding affinity (ka z 1010/M) which could also account for some of the observed differences. The contribution of binding affinity to MB acoustic behavior has not been explored and will be a focus of a future investigation. The local MB confining conditions are also known to have a dramatic effect on the resulting bubble dynamics, wall stiffness can either increase or decrease the resonance frequency of an MB positioned near it (Doinikov et al. 2011; Martynov et al. 2009; Qin and Ferrara 2007; Sassaroli and Hynynen 2005). Rigid
boundaries have the effect of decreasing the resonance frequency, whereas compliant boundaries have the opposite effect. Furthermore, the relative size of the vessel compared with the MB has also been shown to enhance this change in resonance frequency. In this study, the MBs are confined in a 200-mm cellulose capillary fiber compared with the Overvelde study, which used the Opticell culture dishes (polystyrene membranes with 2 mm of clearance). A number of size-dependent effects become more pronounced because the materials are different, and hence the material properties and effects on MB dynamic characteristics, and the capillary fiber is much closer in size to the MBs. One must also examine what it is that the studies are actually gathering as their data sets. The present study examines the scattered pressure and specifically the second harmonic signal generation as an indicator of resonance. In comparison, the Overvelde study is examining the radial oscillations and specifically the maximum radial excursion. By nature, this is predominately a measure of the fundamental MB response. As a further complication, these data are obtained via a high-speed camera limited to a single orthogonal view. As previously stated, an MB near a boundary can experience significant asymmetry in oscillation. Although it has been demonstrated that it is possible to derive the pressure-time trace from the radius-time curve by assuming a purely radial oscillation (Sijl et al. 2011), it has also been shown that this relationship can break down when examining higher-frequency components above the fundamental response. MB response to repeated insonation Figure 9 indicates that resonance plays an important role in the stability and longevity of MBs. This stands to
Characterization of adherent microbubbles d J. CASEY et al.
reason when one considers that a bubble at resonance will experience the widest fluctuations in size and shape deformations. This finding is supported by Guidi et al. (2010) and Vos et al. (2009), who used a high-speed camera approach to show that phospholipid-shelled MBs undergo acoustically induced deflation after repeated insonation and that this is exacerbated near resonance. The shift in resonance size shown in the second harmonic scattering is also clearly evident in both parts of Figure 9 by the shift in disruption threshold to larger MBs for adherent MBs. Unlike the second harmonic generation, however, there is no distinct disruption peak associated with the resonance frequency. The stability index indicates that bubbles smaller than the resonance radius undergo severe disruption even at the relatively low acoustic pressure (specifically the small radii MBs of Fig. 9); the caveat is that this is true only for these specific experimental conditions. It could be that given larger MBs and a lower driving frequency, the bubbles smaller than the resonance radius might be robust enough to survive. The use of the broadband destructive pulse activated a wider size distribution of MBs as demonstrated by the increase in disruption threshold for both adherent and non-adherent MBs. In the future, however, it might be useful to consider using either a broader band or a transducer transmitting a lower center frequency for the activation of a more diverse population of the MBs. With the current setup, it can be seen that the majority of MBs interrogated are still larger than the size range affected by the ultrasound in both the second harmonic and stability index data. A number of MBs show that there has been an increase in the total scattering to successive pulse (a stability index .1). As already described, MBs are known to exhibit acoustically induced deflation in response to ultrasound (Guidi et al. 2010). By this mechanism, if an MB with a radius initially greater than the resonance radius was insonated, each successive pulse could deflate the MB, driving its radius toward the resonance size. In this case and with reference to Figure 8, it can be seen that for little change in fundamental scattering as the MB approaches resonance, the second harmonic contribution to total scattering would result in an increase in total scattering and hence a value greater than 1 for the stability index. Directions for future investigation The shift in acoustic frequency detected could have potential for the development of novel imaging strategies for the differentiation of adherent MBs. At present, the detected shift in resonance frequency is small and subject to variation between MBs (even of similar size). In most bulk or non-single bubble techniques, these subtle differ-
913
ences will be masked by the response from the polydisperse MB population. Currently there is a significant effort developing methods for the production of monodisperse populations of MBs (Talu 2007; Talu et al. 2006; Xu et al. 2006). With tightly controlled MB physical characteristics (e.g., radius, shell composition), the variability between MBs of similar sizes will be reduced and the MB radius can be tailored to the specific imaging parameters and make maximal use of any resonance frequency shift. Furthermore, the examination of sub-harmonics (SHs) rather than higher harmonics might be worth considering. SHs are produced at half the insonation frequency and have been shown to be at a maximum when an MB is driven at twice its resonance frequency (Katiyar and Sarkar 2010). If the shift in resonance observed here is also evident in the SH response of MBs, this would effectively result in the difference in resonance radius being exaggerated twofold, allowing for easier differentiation. CONCLUSIONS The acoustic setup for detailed interrogation of single MBs under controlled conditions is presented. The adherence of MBs to a capillary fiber wall via a biotin–streptavidin bond can be seen to increase the MB resonance frequency by shifting the radius of peak generation of second harmonic signals. The fundamental scattering properties of the MBs are unchanged by the presence of adhesion. MB stability is heavily dependent on the resonance properties, with resonance bubbles being significantly more susceptible to disruption than MBs situated away from the resonance radius. The change in resonance frequency shown by the second harmonic generation is also highlighted by the bubble disruption. Acknowledgments—The authors would like to acknowledge the contributions from Jonathan Loughran, Dr. Richard Browning and Dr. Emma Behjat. This work was supported by Engineering and Physical Sciences Research Council grant number EP/G038163/1.
REFERENCES Anderson DR, Tsutsui JM, Xie F, Radio SJ, Porter TR. The role of complement in the adherence of microbubbles to dysfunctional arterial endothelium and atherosclerotic plaque. Cardiovasc Res 2007; 73:597–606. Borden MA, Pu G, Runner GJ, Longo ML. Surface phase behavior and microstructure of lipid/PEG-emulsifier monolayer-coated microbubbles. Colloids Surf B Biointerfaces 2004;35:209–223. Caskey CF, Kruse DE, Dayton PA, Kitano TK, Ferrara KW. Microbubble oscillation in tubes with diameters of 12, 25, and 195 microns. Appl Phys Lett 2006;88:033902. Caskey CF, Stieger SM, Qin S, Dayton PA, Ferrara KW. Direct observations of ultrasound microbubble contrast agent interaction with the microvessel wall. J Acoust Soc Am 2007;122:1191–1200. Chomas JE, Dayton P, May D, Ferrara K. Threshold of fragmentation for ultrasonic contrast agents. J Biomed Opt 2001;6:141–150.
914
Ultrasound in Medicine and Biology
Couture O, Bannouf S, Montaldo G, Aubry J, Fink M, Tanter M. Ultrafast imaging of ultrasound contrast agents. Ultrasound Med Biol 2009;35:1908–1916. Dayton JJR. Molecular ultrasound imaging using microbubble contrast agents. Front Biosci 2007;12:5124–5142. Doinikov A, Leila A, Ayache B. Acoustic scattering from a contrast agent microbubble near an elastic wall of finite thickness. Phys Med Biol 2011;56:6951. Doinikov AA, Haac JF, Dayton PA. Resonance frequencies of lipidshelled microbubbles in the regime of nonlinear oscillations. Ultrasonics 2009;49:263–268. Dollet B, van der Meer SM, Garbin V, de Jong N, Lohse D, Versluis M. Nonspherical oscillations of ultrasound contrast agent microbubbles. Ultrasound Med Biol 2008;34:1465–1473. Frinking PJA, Gaud E, Brochot J, Arditi M. Subharmonic scattering of phospholipid-shell microbubbles at low acoustic pressure amplitudes. IEEE Trans Ultrason Ferroelectr Freq Control 2010;57: 1762–1771. Garbin V, Cojoc D, Ferrari E, Di Fabrizio E, Overvelde MLJ, van der Meer SM, de Jong N, Lohse D, Versluis M. Changes in microbubble dynamics near a boundary revealed by combined optical micromanipulation and high-speed imaging. Appl Phys Lett 2007; 90:114103. 3. Guidi F, Tortoli P, Vos HJ, de Jong N. Simultaneous optical and acoustical observation of microbubbles behaviour. IEEE Ultrasonics Symposium, Vancouver, Canada, 2006. Guidi F, Vos HJ, Mori R, De Jong N, Tortoli P. Microbubble characterization through acoustically induced deflation. IEEE Trans Ultrason Ferroelectr Freq Control 2010;57:193–202. Helfield BL, Cherin E, Foster FS, Goertz DE. Investigating the subharmonic response of individual phospholipid encapsulated microbubbles at high frequencies: A comparative study of five agents. Ultrasound Med Biol 2012;38:846–863. Katiyar A, Sarkar K. Excitation thresholds for subharmonic response of ultrasound contrast microbubbles. J Acoust Soc Am 2011;130: 3137–3147. Lindner JR. Assessment of inflammation with contrast ultrasound. Prog Cardiovasc Dis 2001;44:111–120. Lindner JR. Microbubbles in medical imaging: current applications and future directions. Nat Rev Drug Discov 2004;3:527–533. Lindner JR, Dayton PA, Coggins MP, Ley K, Song J, Ferrara K, Kaul S. Noninvasive imaging of inflammation by ultrasound detection of phagocytosed microbubbles. Circulation 2000;102:531–538. Marmottant P, van der Meer S, Emmer M, Versluis M, de Jong N, Hilgenfeldt S, Lohse D. A model for large amplitude oscillations of coated bubbles accounting for buckling and rupture. J Acoust Soc Am 2005;118:3499–3505. Martynov S, Stride E, Saffari N. The natural frequencies of microbubble oscillation in elastic vessels. J Acoust Soc Am 2009;126:2963–2972. Morgan KE, Allen JS, Dayton PA, Chomas JE, Klibaov AL, Ferrara KW. Experimental and theoretical evaluation of microbubble behavior: effect of transmitted phase and bubble size. IEEE Trans Ultrason Ferroelectr Freq Control 2000;47:1494–1509. Mulvana H, Stride E, Tang M-X, Hajnal JV, Eckersley RJ. The influence of gas saturation on microbubble stability. Ultrasound Med Biol 2012;38:1097–1100. Needles A, Couture O, Foster FS. A method for differentiating targeted microbubbles in real time using subharmonic micro-ultrasound and interframe filtering. Ultrasound Med Biol 2009;35:1564–1573. Ohl SW, Klaseboer E, Khoo BC. The dynamics of a non-equilibrium bubble near bio-materials. Phys Med Biol 2009;54:6313–6336. Overvelde M, Garbin V, Dollet B, de Jong N, Lohse D, Versluis M. Dynamics of coated microbubbles adherent to a Wall. Ultrasound Med Biol 2011;37:1500–1508. Overvelde M, Garbin V, Sijl J, Dollet B, de Jong N, Lohse D, Versluis M. Nonlinear shell behavior of phospholipid-coated microbubbles. Ultrasound Med Biol 2010;36:2080–2092. Overvelde M. PhD Thesis. Ultrasound Contrast Agents Dynamics of Coated Bubbles, University of Twente, Netherlands, 2010.
Volume 39, Number 5, 2013 Patil AV. A 3D nonlinear FEA model for simulation of microbubble behavior. IEEE International Ultrasonics Symposium Proceedings, Beijing, China, 2008. Pitt WG. Ultrasonic drug delivery—A general review. Expert opin Drug Deliv 2004;1:37–56. Qin S, Ferrara KW. The natural frequency of nonlinear oscillation of ultrasound contrast agents in microvessels. Ultrasound Med Biol 2007;33:1140–1148. Sassaroli E, Hynynen K. Resonance frequency of microbubbles in small blood vessels: A numerical study. Phys Med Biol 2005;50: 5293–5305. Sboros V, Glynos E, Pye SD, Moran CM, Butler M, Ross J, Short R, McDicken WN, Koutsos V. Nanointerrogation of ultrasonic contrast agent microbubbles using atomic force microscopy. Ultrasound Med Biol 2006;32:579–585. Sboros V, Moran CM, Pye SD, McDicken WN. The behaviour of individual contrast agent microbubbles. Ultrasound Med Biol 2003;29: 687–694. Sboros V, Pye SD, MacDonald CA, Gomatam J, Moran CM, McDicken WN. Absolute measurement of ultrasonic backscatter from single microbubbles. Ultrasound Med Biol 2005;31:1063–1072. Sennoga CA, Mahue V, Loughran J, Casey J, Seddon JM, Tang M, Eckersley RJ. On sizing and counting of microbubbles using optical microscopy. Ultrasound Med Biol 2010;36:2093–2096. Sijl J, Vos HJ, Rozendal T, de Jong N, Lohse D, Versluis M. Combined optical and acoustical detection of single microbubble dynamics. J Acoust Soc Am 2011;130:3271–3281. Talu E. Tailoring the size distribution of ultrasound contrast agents: Possible method for improving sensitivity in molecular imaging. Mol Imaging 2007;6:384–392. Talu E, Hettiarachchi K, Nguyen H, Lee AP, Powell RL, Longo ML, Dayton PA. Lipid-stabilized monodisperse microbubbles produced by flow focusing for use as ultrasound contrast agents. Proceedings of the IEEE Ultrasonics Symposium, Vancouver, Canada, 2006. Tu J, Guan J, Qiu Y, Matula TJ. Estimating the shell parameters of SonoVue microbubbles using light scattering. J Acoust Soc Am 2009; 126:2954–2962. Tu J, Swalwell JE, Giraud D, Cui W, Chen W, Matula TJ. Microbubble sizing and shell characterization using flow cytometry. IEEE Trans Ultrason Ferroelectr Freq Control 2011;58:955–963. van der Meer SM, Versluis M, Lohse D, Chin CT, Bouakaz A, de Jong N. The resonance frequency of SonoVue; as observed by high-speed optical imaging. Ultrasonics Symposium, 2004 IEEE, Montreal, Canada, 2004. pp. 343–345 Vol. 1. Vos HJ, de Jong N, Mori R, Viti J, Guidi F, Tortoli P, Sijl J. Ultrafast framing optical and acoustical recording of nonlinearly oscillating microbubbles. Ultrasonics Symposium (IUS), 2009 IEEE International, Rome, Italy, 2009. pp. 1817–1820. Vos HJ, Dollet B, Bosch JG, Versluis M, de Jong N. Nonspherical vibrations of microbubbles in contact with a wall–a pilot study at low mechanical index. Ultrasound Med Biology 2008;34:685–688. Wijkstra H, Smeenge M, Rosette J, Pochon S, Tardy-Cantalupi I, Tranquart F. Targeted microbubble prostate cancer imaging with BR55. 17th European Symposium on Ultrasound Contrast Imaging, Rotterdam, Netherlands, 2012. Willmann JK, Paulmurugan R, Chen K, Gheysens O, Rodriguez-Porcel M, Lutz AM, Chen IY, Chen X, Gambhir SS. US imaging of tumor angiogenesis with microbubbles targeted to vascular endothelial growth factor receptor type 2 in Mice1. Radiology 2008;246:508–518. Xie F, Tsutsui JM, Lof J, Unger EC, Johanning J, Culp WC, Matsunaga T, Porter TR. Effectiveness of lipid microbubbles and ultrasound in declotting thrombosis. Ultrasound Med Biol 2005; 31:979–985. Xu JH, Li SW, Chen GG, Luo GS. Formation of monodisperse microbubbles in a microfluidic device. AIChE J 2006;52:2254–2259. Zhao S. Detection of echoes from adherent targeted microbubbles. IEEE Ultrasonics Symposium, Vancouver, Canada, 2006. Zhao S, Kruse DE, Ferrara KW, Dayton PA. Selective imaging of adherent targeted ultrasound contrast agents. Phys Med Biol 2007; 52:2055–2072.