Ultrasound in Med. & Biol., Vol. 45, No. 8, pp.

21882204, 2019
Copyright © 2019 The Author(s). Published by Elsevier Inc. on behalf of World Federation for Ultrasound in Medicine & Biology.
This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/)
Printed in the USA. All rights reserved.
0301-5629/$ - see front matter

https://doi.org/10.1016/j.ultrasmedbio.2019.04.005

Original Contribution

NON-LINEAR ACOUSTIC EMISSIONS FROM THERAPEUTICALLY DRIVEN

CONTRAST AGENT MICROBUBBLES

TAGEDPJAE HEE SONG,* ALEXANDRU MOLDOVAN,y and PAUL PRENTICE*TAGEDEN

* CavLab, Medical and Industrial Ultrasonics, University of Glasgow, Glasgow, United Kingdom; and y Centre for Ultrasound
Engineering, University of Strathclyde, Glasgow, United Kingdom
(Received 29 August 2018; revised 25 March 2019; in final from 1 April 2019)

Abstract—Non-linear emissions from microbubbles introduced to the vasculature for exposure to focused ultra-
sound are routinely monitored for assessment of therapy and avoidance of irreversible tissue damage. Yet the
bubble-based mechanistic source for these emissions, under subresonant driving at typical therapeutic pressure
amplitudes, may not be well understood. In the study described here, dual-perspective high-speed imaging at
210,000 frames per second (fps), and shadowgraphically at 10 Mfps, was used to observe cavitation from micro-
bubbles flowing through a 500-mm polycarbonate capillary exposed to focused ultrasound of 692 kHz at thera-
peutically relevant pressure amplitudes. The acoustic emissions were simultaneously collected via a broadband
calibrated needle hydrophone system. The observations indicate that periodic bubble-collapse shock waves can
dominate the non-linear acoustic emissions, including subharmonics at higher driving amplitudes. Contributions
to broadband emissions through variance in shock wave amplitude and emission timings are also identified. Pos-
sible implications for in vivo microbubble cavitation detection, mechanisms of therapy and the conventional clas-
sification of cavitation activity as stable or inertial are discussed. (E-mail: [email protected].
uk) © 2019 The Author(s). Published by Elsevier Inc. on behalf of World Federation for Ultrasound in
Medicine & Biology. This is an open access article under the CC BY license.
(http://creativecommons.org/licenses/by/4.0/).
Key Words: Microbubble, Cavitation, Focused ultrasound, Acoustic emissions, Bubble collapse, Shock wave.

INTRODUCTION various combinations of harmonic emissions at nf0

(where n is any integer), subharmonics at nf 0/2 and
During the application of focused ultrasound for tissue
broadband emissions are commonly reported. Generally
disruption and drug delivery, contrast agent microbubble
for therapeutic applications, harmonic emissions are
suspensions are often injected into the vasculature to
associated with lower driving amplitudes, with higher
mediate therapy (Ferrara et al. 2007). Non-linear acous-
threshold values required for subharmonic and broad-
tic emissions generated by the driven microbubbles are
band emissions (Jones et al. 2018; McDannold et al.
routinely monitored for assessment of therapeutic bio-
2006). Indeed, rapid control feedback loops modulating
effect. This is particularly the case for applications in
the driving amplitude in response to non-linear emission
which avoidance of overtreatment is critical, such as
components (O’Reilly and Hynynen 2012; Sun et al.
transcranial bloodbrain barrier disruption, for which
2017) may be an important technology for safe clinical
recent clinical (Lipsman et al. 2018) and extensive pre-
realisation of microbubble-mediated therapy of the
clinical (Jones et al. 2018; O’Reilly and Hynynen 2012;
brain.
Sun et al. 2017; Wu et al. 2014) development is under-
A volume of literature exists detailing direct obser-
way. With typical focused ultrasound fundamental driv-
vation of microbubbles via high-speed optical imaging,
ing frequencies (f0) of several hundreds of kilohertz
under higher driving frequencies typical of diagnostic
employed, for sufficient transmission across the skull,
imaging, close to or above microbubble resonance. This
research aims to identify sources of non-linear emissions
Address correspondence to: Paul Prentice, Cavitation Research responsible for contrast enhancement, the original
Laboratory, James Watt South Building, School of Engineering,
University of Glasgow, Glasgow G12 8QQ, UK. E-mail:
application for microbubble suspensions in medical
[email protected] ultrasound. A range of microbubble behaviour such as

2188
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2189

compression-only (de Jong et al. 2007; Sijl et al. 2011) generator (DG4102, Rigol Technologies, Beijing, China)
and non-spherical oscillations (Dollet et al. 2008; Versluis and a power amplifier (2100 L, Electronics and Innova-
et al. 2010) have been proposed as sources of non-linear tion, Rochester, NY, USA), was mounted in a chamber
emissions, including spherical period-doubling (Chomas on an xyz manipulator (Velmex Motor, Bloomfield, NY,
et al. 2002) and subharmonic (Sijl et al. 2010) responses USA). The transducer has an outer diameter of 110 mm,
for driving at around twice microbubble resonance. These and geometrically focuses to 68 mm from the front face.
studies, however, lacked parallel acoustic monitoring for The field generated propagates horizontally in the
verification that the identified behaviour translated to a x-direction of a chamber, measuring 420 £ 438 £ 220
detectable component within the acoustic emissions. mm3, that also features two recessed opposing walls
Other reports on high-speed imaging of microbub- (along the z-axis), to allow positioning of optics for rea-
bles, driven at lower frequencies (
1 MHz) and pressure sonably high spatial resolution imaging (side view,
amplitudes typical of those used for therapy (Caskey described below). The chamber was filled with de-ion-
et al. 2007; Chen et al. 2011a, 2011b; Lajoine et al. ised water, degassed via boiling and cooling in sealed
2018; Prentice et al. 2005; van Wamel et al. 2006), seek containers, to <4 mg/L dissolved oxygen.
primarily to identify mechanisms of tissue disruption A polycarbonate capillary (inner/outer diameter
and drug delivery. Potential mechanistic sources for 500 mm/550 mm, Paradigm Optics, Vancouver, WA,
non-linear emissions under such driving conditions, USA) was mounted into the chamber via a custom 3-D
however, are not so well addressed. Clearly, a robust printed support, such that the capillary was located at the
understanding of the source of non-linear emissions acoustic focus, at 45˚ to the propagation axis and in the
from microbubbles driven below resonance would aid horizontal plane (Fig. 1a). PNP amplitude measurements
detection and monitoring of therapy, and may also pro- at the focus with the 0.2-mm polyvinyl difluoride needle
vide insights into the mechanisms underpinning tissue hydrophone (fully described below, but for focused
disruption and drug delivery. ultrasound characterisation measurements, aligned to the
Here, we report on observations from diluted sam- axis of propagation), with and without the capillary in
ples of SonoVue contrast agent microbubbles flowing position, indicated attenuation of <5% across the PNPs
through a 500-mm polycarbonate capillary, driven by a stated in the Results, with free-field measurements
200-cycle burst of 692-kHz focused ultrasound, at peak- reported. All observations were taken over a 200-cycle
negative pressure (PNP) amplitudes ranging from burst of 692-kHz focused ultrasound (»289 ms), operat-
3851.14 MPa. The capillary is mounted at 45˚ relative ing at the third harmonic of the transducer, with initial
to the axis of focused ultrasound propagation, to facili- excitation denoting t = 0 ms. Use of the third harmonic
tate dual high-speed imaging from orthogonal perspec- also introduced a transition response for the focused
tives, represented schematically in Figure 1a. Top-view ultrasound generated to ramp up to the required PNP
imaging is recorded at 210,000 frames per second (fps) amplitude, with measurements indicating »16 ms to
for the duration of the focused ultrasound exposure. achieve 90% of the value stated. The significant har-
Side-view imaging is captured at 10 Mfps for a duration monic components from non-linear propagation of the
of 25.6 ms, from »65 ms after focused ultrasound has focused ultrasound, at each PNP reported, are illustrated
propagated to the capillary. Side-view illumination is in Figure 1b and summarized in Table 1, for 459 kPa.
provided by collimated synchronous 10-ns laser pulses, SonoVue (Bracco, Milan Italy) is a commercial
for shadowgraphic imaging of acoustic transients as they ultrasound contrast agent consisting of phospholipid-
propagate through the field of view (FOV). Acoustic shelled sulphur hexafluoride gas-core microbubbles,
emissions from the imaged microbubble cavitation are with a mean diameter between 2 and 3 mm (95%
sampled via a broadband calibrated needle hydrophone <10 mm), and associated resonance frequencies
system, with the sensing tip located within 100 mm from >2 MHz, dependent on initial microbubble diameter
the capillary outer wall, such that it is visible in both (van der Meer et al. 2004). SonoVue suspensions at an
imaging FOVs. Acoustic data are presented in time and initial concentration of 100500 million/mL (Schneider
frequency domain, deconvolved from the impulse 1999) were prepared according to manufacturer’s guid-
response of the hydrophone over a representative band- ance and subsequently diluted by a factor of »1:80,000
width and filtered, as described below. in (non-degassed) de-ionised water, using a syringe. A
new phial of SonoVue was used for each day of experi-
mental acquisition, with fresh diluted samples prepared
METHODS
from it on an hourly basis. A series of preliminary
Experimental configuration experiments were also conducted using saline as the
A focused ultrasound transducer (H-149, Sonic dilution medium. Microbubble cavitation activity obser-
Concepts, Bothell, WA, USA), excited by a waveform vations equivalent to those presented in the Results
2190 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

Fig. 1. (a) Schematic representation of the experimental configuration, depicting exposure of the capillary, orientated in
the horizontal plane at 45˚ to the focused ultrasound, and the needle hydrophone in the emission collection position.
Focused ultrasound propagates from left to right, and microbubble flow was from back right to front left, for this repre-
sentation. The imaging axis and fields of view for the high-speed imaging are also represented (top view dotted, side
view dashed). (b) Assessment of focused ultrasound propagation non-linearity, up to 4f0 for the peak-negative pressure
amplitudes reported in the Results.

(according to PNP) suggested that microbubbles are rea- focused ultrasound exposure. At this frame rate, images
sonably stable, in water, over this time scale. The syringe are formed over 384 £ 160 pixels with a spatial resolu-
was then connected to a 20G microlance (»0.6-mm tion of 4.2 mm/pixel, which is insufficient to resolve the
internal diameter), with one end of the capillary epoxied majority of quiescent microbubbles, before focused
within the bore of the lance and mounted into a syringe ultrasound excitation. The FOV, represented by the
pump (74900, Cole Parmer, Cambridgeshire UK), set to dotted rectangle in Figure 1a, however, covers the width
fixed flow rate of 7.5 mL/h. The exhaust end of the capil- and »1.5 mm of the length of the capillary within the
lary was vented to a collection reservoir located outside acoustic focus, to ensure that the needle hydrophone, in
the chamber. the position represented in Figure 1a, detected only
emissions generated by imaged microbubble cavitation,
Dual high-speed imaging during the focused ultrasound exposure. Before, and during
High-speed imaging was conducted along the verti- data acquisition, the vertical positioning of the objective
cal y-axis (top-view) and horizontal z-axis (side-view), lens was adjusted to obtain focus on the capillary walls at
orthogonally to the axis of focused ultrasound propaga- their widest point. Illumination for top-view imaging was
tion. Top-view imaging was captured through a long- provided by a continuous 150-W halogen bulb light source
working-distance objective lens (5 £ 0.14 Numerical (Thorlabs, Ely UK), coupled to a liquid light guide, with
Aperture (NA), focal length (in air): 40.0 mm Mitutoyo, the output end located »10 mm below the capillary (not
Kawasaki Japan) in a sealed watertight casing, with a shown, Fig. 1a). Figure 2c depicts 7.5 § 0.2-mm-diameter
Fastcam SA-Z 2100 K (Photron, Bucks UK) at 210,000 polymer microspheres (PS06005, Bangs Laboratories, IN,
fps and shutter time of 159 ns, for the duration of the USA), imaged through the top-view configuration at full
FOV, for further indication of resolution.
Table 1. Relative magnitudes of focused ultrasound non-linear Side-view imaging, with the FOV represented as a
propagation harmonics at 459 kPa (Fig. 1b) and spectral har- dashed rectangle in Figure 1a, was undertaken through a
monic components from microbubble-cavitation driven at a Monozoom 7 lens (Bausch & Lomb, Rochester, NY,
PNP of 459 kPa (Fig. 4c, dark grey) USA) at 10 Mfps with a Shimadzu HPV-X2 camera
Harmonic magnitude (dB) (Shimadzu, Kyoto Japan) over a 25.6-ms duration from
t = 125 ms, such that microbubble cavitation has evolved
f0 2f0 3f0 4f0
over »45 cycles of focused ultrasound driving. Illumina-
Focused ultrasound at PNP = 459 kPa 101.1 64.82 37.75 31.79 tion was achieved with synchronous (to frame capture)
Microbubble cavitation emission spectrum 136.5 105.3 91.92 89.95 10-ns laser pulses (CAVILUX Smart, Cavitar, Tampere,
PNP = peak-negative pressure. Finland), coupled to a liquid light guide and collimating
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2191

Fig. 2. Representative images extracted from a top-view sequence captured at 210 kfps, from t = 0 ms for the duration of
the focused ultrasound exposure, of microbubble cavitation activity driven by peak-negative pressures of (a) 459 kPa
and (b) 1.14 MPa. Scale is provided by the 500-mm internal diameter of the capillary, flow is from left to right and the
tip of the needle hydrophone (arrowed NH) is visible to the top left of each image, with the axis of the needle represented
(white dashed line, second frame). Focused ultrasound propagation direction is depicted at 161.9 ms in (a), with trigger-
ing of focused ultrasound defined as t = 0 ms, and propagation time to the capillary = 51.6 ms. (c) 7.5-mm-diameter poly-
mer microspheres, under equivalent imaging conditions and flow conditions, for indication of resolution.

lens. This configuration facilitates shadowgraphic imag- 100 kHz20 MHz in 25-kHz increments (National Physical
ing such that pressure transients can be directly visual- Laboratory, Teddington, UK, November 2017). There is a
ised via refractive index variations imposed as the §9% magnitude uncertainty associated with the calibration,
transient propagates. Note that the duration of side-view across the frequency ranges reported, including the meas-
capture is apparent in the full-sequence videos of top- urements of focused ultrasound PNP amplitude. Voltage
view imaging, as the pulsed laser illumination reflects data from the needle hydrophone system were amplified by
off the bevelled tip of the needle hydrophone. Empiri- »25 dB (Hydrophone Amplifier, Precision Acoustics) and
cally, we have found that the optimal plane of focus for collected, for the duration of the focused ultrasound expo-
imaging acoustic transients (Kudo 2015), whilst retain- sure, to an oscilloscope (MS07104 A, Agilent Technolo-
ing a perception of cavitation dynamics, is achievable by gies, Lexington, MA, USA), at 4 Giga-samples (GS)/s.
defocusing the monozoom lens by »1.5 mm, such that Triggering and synchronisation of all instrumentation was
the bubble itself is slightly out of focus during side-view achieved with a delay generator (DG535, Stanford Research
capture (Song et al. 2016). The FOV for this perspective Systems, Sunnyvale, CA, USA).
was selected to ensure all bubble-collapse shock waves During the experiments, the needle hydrophone was
from microbubble cavitation were captured before prop- mounted on an xyz micro-manipulator (M-652, New-
agating beyond the imaged region, according to a 0.1-ms port, Oxfordshire UK) such that the tip could be posi-
frame-to-frame interval. We also note that the collimated tioned »100 mm from the outer capillary wall, with the
laser illumination introduces a limb effect to the imaging needle shaft »15˚ from orthogonal to the capillary axis
of the capillary, whereby the capillary wall appears and »20˚ below the horizontal plane. This angle of
much larger than the physical 25-mm thickness (see Sup- approach was necessary to facilitate imaging from the
plementary videos S3 and S4, at full FOV). Although two perspectives described and because of the physical
smaller-internal-diameter capillaries are available, they constraint of the recessed side wall of the chamber archi-
tend to have similar wall thicknesses such that the limb tecture. The needle hydrophone tip effectively defined
effect would obscure microbubble cavitation imaging. the target region of interest for detectable microbubble
activity, within the capillary, on incidence of focused
Acoustic detection ultrasound.
The acoustic emissions from the driven microbubble All single microbubble cavitation time domain data
cavitation activity were sampled with a 0.2-mm polyvinyl are presented from »60175 ms, with the needle hydro-
difluoride needle hydrophone (Precisions Acoustics, Dor- phone (in field characterisation position, with the tip
chester, UK), with sensitivity and phase calibration from located at the capillary position), indicating a propagation
2192 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

time from the transducer of 51.6 ms. The impulse of the SonoVue population (van der Meer et al. 2004).
response of the needle hydrophone has been deconvolved We also note that we occasionally observed larger
from the voltage data over a selected bandwidth of 2.4 microbubbles flowing through the capillary, immediately
(»7f0/2) to 20 MHz. This removes the fundamental fre- after sample preparation, during “live” imaging. For
quency of the focused ultrasound and significant non- high-speed imaging acquisition, however, cavitation was
linear components, for revealing shock wave content and always initiated from unresolved microbubbles. We pos-
restoring an approximation to the physical pressure data tulate that buoyancy within the syringe prevented larger
(Johansen et al. 2017). Frequency spectra are generated microbubbles from entering the capillary, soon after sam-
via application of a fast Fourier transform and Hanning ple preparation. Because of the angle of focused ultra-
window to the time interval presented, but deconvolved sound incidence to the capillary (Fig. 1a), the primary
across the full calibration bandwidth of 100 radiation force of the exposure acts to translate the micro-
kHz20 MHz, and thereby include the focused ultra- bubble cavitation activity at »45˚, relative to the axis of
sound driving as well as microbubble cavitation emission the capillary, and in the general direction of the needle
components. hydrophone tip. The capillary wall distal to the trans-
The experiment described was repeated at each ducer, however, may also be expected to influence the tra-
focused ultrasound PNP amplitude reported, until data jectory of cavitation translation as it is approached.
were recorded from a single microbubble within the tar- Within the limit of the temporal resolution of the
get region of the capillary, defined by the tip of the nee- top-view imaging (see Methods), the microbubble driven
dle hydrophone, and with the ensuing cavitation activity at a PNP of 459 kPa shows no indication of fragmenta-
reasonably in focus for the top-view imaging (close to tion and translates with an average speed of » 3.4 m/s
the central horizontal plane of the capillary). We note (sampled between 70150 ms), into contact with the dis-
that needle hydrophone detection of non-linear signals, tal capillary wall at »152 ms. No further bubble activity
other than those attributable to the focused ultrasound is observed for the rest of the imaging sequence after
propagation (Fig. 1b), were consistently accompanied by contact, although it is possible that curvature of the cap-
observable microbubble cavitation activity in the high- illary wall obscures any bubble from view. In contrast,
speed imaging from both perspectives (and vice versa), the microbubble driven by 1.14 MPa fragments within
initiating from focused ultrasound incidence to the capil- the first few cycles of driving. Bubble debris trailing the
lary, at the required PNP. Sham exposures conducted cloud translation, such as that arrowed at 138.1 ms, was
regularly during data collection, for which microbubbles commonly observed from both imaging perspectives, at
were not introduced to the flow, and at PNP values in higher PNP amplitudes. Cloud translation is at a reduced
excess of those at which results are presented below, average speed of »1.1 m/s, possibly because of higher
confirmed that no cavitation activity was observed, or drag forces acting during the larger inflation phases
acoustically detected, in the absence of microbubbles. (Johnston et al. 2014), until contact with the distal capil-
lary wall at »280 ms. Subsequent translation continues
along the inner capillary wall, until the end of the
RESULTS
focused ultrasound burst.
f0 and f0/2 shock wave emission regimes In Figure 3(a, b) are representative images from the
In Figure 2 are representative top-view images of side-view shadowgraphic sequence, over two consecu-
single-microbubble cavitation behaviour, driven with tive cycles of driving, of the bubble activity captured
focused ultrasound of PNPs (a) 459 kPa and (b) 1.14 from top-view in Figure 2(a, b), respectively. These
MPa (full image sequence at full FOV, available as Sup- images (and those of the full sequence, over »17 cycles,
plementary videos S1 and S2, respectively, online only). at full FOV; Supplementary Videos S3 and S4, online
Images up to and including the focused ultrasound inci- only) have been background-subtracted, with a sequence
dence to the capillary, at 51.6 ms, indicate that no micro- recorded immediately after the bubble activity was cap-
bubbles of initial quiescent size larger than the spatial tured, but with no focused ultrasound generated, for the
resolution of this perspective are flowing within the cap- purpose of enhancing shadowgraphic features.
illary. It is only once the pressure amplitude has ramped Figure 3a, of the microbubble driven at 459 kPa, further
up sufficiently, during the »16-ms transition response suggests that the resulting cavitation initiated from a sin-
associated with driving the transducer at its third har- gle microbubble, and did not fragment, with apparently
monic (see Methods), that the microbubble cavitation radial oscillations in phase with the driving. Figure 3b
becomes observable in the imaging, at (a) 71.4 ms and illustrates the behaviour of the cavitation cloud resulting
(b) 61.9 ms for PNP = 459 kPa and 1.14 MPa, respec- from the microbubble driven by a PNP of 1.14 MPa.
tively. This suggests that the focused ultrasound induces The pressure phases of the focused ultrasound driving at
cavitation from subresonant microbubbles, likely typical the acoustic focus are apparent from the background
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2193

Fig. 3. Images extracted from the side-view shadowgraphic sequence at 10 Mfps, from t = 125 ms, of the microbubble
cavitation activity also represented from top-view imaging in Figure 2. Flow is from right to left and out of plane. Images
are background-subtracted to enhance shadowgraphic features, including focused ultrasound driving and bubble collapse
shock waves, arrowed red (b). Bar = 150 mm; however, the cavitation activity appears larger as the imaging is purpose-
fully defocused for shadowgraphic capability.

brightness, with rarefactions (133.8 and 135.2 ms) present- focused ultrasound exposure continuing up to »350 ms.
ing as brighter, and compressions (134.6 ms) as darker. The spectrum of the signal, dark grey in Figure 4c, is
This effect is particularly prominent in the video version hydrophone deconvolved over the calibration bandwidth
of the full image sequence (Supplementary Video S4, (100 kHz20 MHz) and therefore inclusive of the
online only). The cloud exhibits overall oscillations at f0 focused ultrasound. It features peaks at nf0 reducing in
of the driving, but only collapses strongly enough to emit magnitude with increasing frequency, and a noise floor
a shock wave for every other compression, with the same at »60 dB, close to instrumental noise for the needle
shock wave arrowed at 132.8 and 132.9 ms, and that from hydrophone system, sampled before focused ultrasound
the next collapse, arrowed at 135.7 ms, » 2/f0 later. The incidence to the capillary (orange dot). Table 1 details
deflation in response to the intermediate compression is the harmonic magnitudes up to 4f0 for the spectrum, rela-
captured at 134.6 ms. The creation of bubble debris after tive to the respective value of the non-linear focused
collapse, usually to the left of the main cloud in accor- ultrasound driving, at 459 kPa (Fig. 1b), indicating the
dance with the direction from which the focused ultra- spectral contribution of f0-periodic shock waves.
sound has propagated, and re-merging with the cloud over The acoustic emission data for the microbubble cavi-
the subsequent cycle are also apparent. tation driven at 1.14 MPa (Fig. 4b) indicates that the bubble
Figure 4(a, b) illustrates the microbubble cavitation initially responds with f0-shock wave emission, including
acoustic emission data recorded by the needle hydro- for a limited duration after fragmentation. Shock waves
phone during the high-speed observations of Figures 1 emitted at f0/2 are apparent from »70 ms, confirming the
and 2, filtered and hydrophone deconvolved to reveal side-view observation of Figure 3b, for emission from
and restore shock wave features, as described under cloud collapse during alternate compressional phases, of
Methods. Figure 4a clearly illustrates periodic shock every other cycle of focused ultrasound.
wave emission at f0, consistent with the observation of The shock waves shadowgraphically imaged in
full oscillations in-phase with the focused ultrasound Figure 3b are also arrowed within the dotted rectangle,
driving (Fig. 3a). The gradual increase in peak-positive which represents the time interval over which side-view
pressure amplitudes of the shock waves, over the first imaging was captured. The shock wave peak-positive
»30 ms of focused ultrasound, may in part be due to the pressure amplitudes are notably both higher and more
primary radiation force-induced translation of the bubble variable than the f0-emitted shock waves of Figure 4a,
toward the needle hydrophone tip (Fig. 1a). Notably, such that no clear trend as the cloud translates toward
shock wave emission ceases at »158 ms, corresponding the needle hydrophone tip is apparent. The spectrum of
to bubble contact with the capillary wall, indicating that the emission signal (Fig. 4 c, blue) contains peaks at nf0/
the bubble does not survive the interaction, despite the 2 for all n and a higher noise floor at »85 dB. Figure 4d
2194 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

Fig. 4. (a, b) Needle hydrophone data, filtered and deconvolved from 2.4 to 20 MHz, to reveal shock wave content
within the microbubble cavitation emission signal, from the activity imaged in Figures 2 and 3 at PNP = 459 kPa and
1.14 MPa, respectively. Black-dash boxes correspond to the duration of the shadowgraphic imaging of Figure 3, with
shock waves arrowed in Figure 3b similarly identified in (b). (c) The spectra of the signal collected, deconvolved from
100 kHz to 20 MHz, thereby including focused ultrasound driving. The needle hydrophone instrumental noise floor is
also depicted (orange dot). (d) Variance in shock wave emission interval times measured from (a) and (b).

graphically illustrates the variance in shock wave inter- the synthetic spectrum technique (Song et al. 2016,
val timings—1.448 § 0.006 ms at 459 kPa and 2.890 § 2017), which facilitates identification of the source of
0.073 ms at 1.14 MPa (average § standard deviation)— spectral features mediated by periodic shock wave char-
over the duration of the focused ultrasound burst, rela- acteristics. Briefly, a simulated bubble-collapse shock
tive to the period of the driving, 1/f0  1.45 ms and 2/f0 wave is modelled via the Gilmore equation (Kreider
 2.90 ms (orange dash). et al. 2011) for a freely collapsing bubble (under ambient
atmospheric pressure, with no applied acoustic driving)
Synthetic spectral analysis in water, of maximum radius Rmax = 50 mm and equilib-
In this section we provide analysis for both micro- rium radius R0 = 4.4 mm, and approximating the emis-
bubble cavitation noise spectra of Figure 4c, adopting sion as a spherically propagating wave. The simulated
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2195

shock wave profile is bandpass filtered from 100 Figure 5(a, b) depicts the amplitudes and timings of
kHz20 MHz, according to the calibration bandwidth of the 1st and 13th simulated shock waves, after the refer-
the needle hydrophone. The resulting profile can then be ence shock wave detected at t = 124.148 ms, for the vari-
manually fitted, in terms of amplitude and timing, rela- ous synthetic shock wave trains constructed to analyse
tive to the experimentally detected shock waves. the f0 emission data, relative to the respective experi-
We note that the R0 of the simulated bubble does mentally detected shock wave (dark grey, solid). The
not directly represent the unknown initial size of the profiles of the experimentally detected shock waves are
SonoVue microbubble, which is unresolved by top-view notably wider than those of the simulated counterparts.
imaging before focused ultrasound excitation (Fig. 2a, This is because the initial microbubble position and the
b). The synthetic signal reconstructions described below, microbubble cavitation trajectory during the focused
of the experimentally detected emission signal, which ultrasound exposure (Fig. 2a) is 90100 mm to the right
can vary significantly over the duration of the focused of the needle hydrophone axis. The shock waves emitted
ultrasound exposures including around fragmentation during this activity are therefore incident from an obli-
events, is reasonably insensitive to the simulated bubble- que angle, resulting in an increased “spreading effect”
collapse parameters, particularly after filtering and over the active element of the hydrophone. The no vari-
within the bandwidths presented. ance, amplitude variance, interval variance and both
A reference shock wave is selected as the first variances simulated shock waves are similar, reflecting
detected within the side-view imaging duration, from the relatively stable amplitudes and timings of the exper-
t = 125 ms, for the f0 and f0/2 shock wave emission imentally detected shock waves emitted in the f0 regime
regimes at PNPs of 459 kPa and 1.14 MPa, within the (Fig. 4a, d). In Figure 5(c, d) are the synthetic spectra for
dashed boxes of Figure 4(a, b), respectively. Various the respective shock wave trains, generated via applica-
synthetic shock wave trains are constructed in the time tion of a fast Fourier transform to each synthetic shock
domain, around the reference shock wave, as described wave train over the duration of experimental detection,
below. A fast Fourier transform is then applied, consis- with the spectra of the experimentally detected signal,
tent with the procedure for generating the spectra of the inclusive of focused ultrasound driving and propagation
experimentally detected emission signals, to produce the harmonics (Fig. 1b and Table 1) also presented. The no
corresponding synthetic spectrum. variance spectrum (Fig. 5d, green dot) exhibits spectral
peaks at nf0, with a 0-dB noise floor, consistent with

No variances (green dot): Simulated shock waves are spectral analysis model for periodic shock waves (Song
assigned a constant peak-positive amplitude, taken as et al. 2016). Amplitude variance, interval variance and
the average over all those experimentally measured both variances spectra also feature nf0 peaks, but
within the stable emission duration for each regime, decreasing in magnitude at higher frequencies, and with
and placed at 1/f0 (1.445 ms) or 2/f0 (2.890 ms) time frequency-dependent noise floors each with a maximum
intervals as appropriate, relative to the experimentally value of »40 dB (Fig. 5c). Instrumental noise (Fig. 5d,
measured reference shock wave timing. brown dot), however, of »55 dB is seen to account for

Amplitude variance (orange dot): Simulated shock the noise floor of the spectrum of the experimentally
waves are placed at the 1/f0 or 2/f0 timings, but detected signal for this regime.
assigned a peak-positive pressure amplitude matching For the f0/2 data, the reference shock wave (as the
that of the nearest experimentally detected shock first visible in Supplementary Video S4) was detected by
wave, thereby illustrating the spectral contribution of the needle hydrophone with the peak-positive pressure
amplitude variance. amplitude occurring at 127.380 ms. Figure 6a reveals the

Interval variance (black dot): Simulated shock waves timings and amplitudes of the 1st simulated shock wave
of average peak-positive pressure amplitude are from the various synthetic shock wave trains, relative to
placed at the timings of experimentally detected the next experimental shock wave after the reference
shock waves, thus demonstrating the effect of interval shock wave, detected at 130.315 ms (blue), therefore rep-
variance in emission timings. resenting an emission interval of 2.935 ms. The peak

Both variances (red dash): Simulated shock waves are amplitudes of the simulated shock waves of the no vari-
placed at the timings of experimentally detected ance and amplitude variance trains are located 0.046 ms
shock waves and assigned a peak-positive pressure earlier, illustrating the slight temporal offset for the
amplitude to match that detected. Comparison of the experimentally detected shock wave from “perfect” 2/f0
both variances spectrum with the no variances, emission timing. Figure 6b is the equivalent representa-
amplitude and interval variance spectra facilitates tion around the 5th experimentally detected shock wave,
identification of the source of all spectral features demonstrating that in this instance, the “perfect” 2/f0
within the experimental data. timings of the no variance and amplitude variance
2196 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

Fig. 5. (a, b) The various synthetic simulated shock wave trains, constructed to analyse the f0-shock wave emission regime,
for the microbubble cavitation activity driven at a peak-negative pressure of 459 kPa, sampled around the 1st and 13th shock
waves after the reference shock wave, detected at 124.148 ms. (c) Amplitude variance and interval variance synthetic spectra,
revealing spectral peaks at nf0, and frequency-dependent noise floors, each with a maximum value of »40 dB. (d) No varian-
ces, instrumental noise, both variances (combining amplitude variance and interval variance) and the experimental spectra.

synthetic shock wave trains locate the simulated shock differences introduced to perfect 2/f0 timings, for match-
waves 0.047 ms after that experimentally detected. The ing to the experimentally detected shock wave timings.
no variance synthetic spectrum (Fig. 6d) of the synthetic The noise floor of both variances (combining amplitude
shock wave train with simulated shock wave profiles at and interval variance) (Fig. 6d, red dash), over the level
regular 2/f0 timings and constant average peak-positive of instrumental noise (brown dot), constitutes the broad-
pressure amplitude exhibits peaks at all nf0/2 with a low band component of the microbubble cavitation emission
noise floor of »15 dB. The amplitude and interval vari- signal, for the f0/2 shock wave regime, driven at
ance spectra (Fig. 6c) also exhibit nf0/2 peaks, but with PNP = 1.14 MPa. The remaining magnitude deficit
notably elevated and frequency-dependent noise floors. across features in both f0 and f0/2 synthetic spectrum
The elevation for interval variance is more pronounced reconstructions may be attributed to shock wave spread-
at higher frequencies due to the relatively small temporal ing effects across the active element of the needle
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2197

Fig. 6. (a, b) Synthetic simulated shock wave trains, for f0/2 data analysis, sampled around the first and fifth shock waves
experimentally detected after the reference shock wave, detected at 127.380 ms. (c) Amplitude variance and interval var-
iance spectra, revealing nf0/2 peaks and the respective contributions to the noise floor. (d) No variances, instrumental
needle hydrophone noise, both variances (combining amplitude variance and interval variance) and the experimental
spectra.

hydrophone tip (Song et al. 2016) and non-linear propa- focused ultrasound burst. Side-view imaging (Fig. 8a; Sup-
gation components of the focused ultrasound at the plementary Video S6, online only), which was recorded
respective PNPs (Fig. 1b). after fragmentation occurred, reveals »f0 -shock wave
emission at 147.4 and 148.8 ms (arrowed red) and some
Microbubble cavitation at intermediate PNPs minor deviation from radially symmetric oscillations.
This section provides experimental data equivalent The microbubble driven by a PNP of 863 kPa
to those presented for the f0 and f0/2 shock wave emission (Fig. 7b; Supplemental Video S7, online only) exhibits a
regimes, for focused ultrasound driving at intermediate higher degree of fragmentation, from earlier in the
PNPs of 593 and 863 kPa. focused ultrasound exposure. As for microbubble cavita-
Figure 7a (and Supplementary Video S5, online only) tion driven at 1.14 MPa, the debris tends to form in the
represents the minimal degree of fragmentation observed wake of the cloud translation and generally re-merges
across all experiments, with the fragmentation event occur- with the cloud within the cycle. Figure 8b and Supple-
ring at »104 ms. Before this, the microbubble appears to mentary Video S8 (online only) indicate that shock
cavitate as a single bubble for the initial »27 cycles of the wave emission occurs at both f0 and f0/2 intervals.
2198 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

Fig 7. Representative images extracted from a top-view imaging sequence captured at 210 kfps, from t = 0 ms for the
duration of the focused ultrasound exposure, of microbubble cavitation activity driven by peak-negative pressures of (a)
593 kPa and (b) 863 kPa. Scale is provided by the 500-mm internal diameter of the capillary. The moment of focused
ultrasound generation at the transducer defines t = 0 ms, with a propagation time to the capillary of 51.6 ms.

The acoustic data in Figure 9(a, b) illustrate the wave interval plot of Figure 9d (red), however, indicates
shock wave emission characteristics of microbubble cav- higher variance in shock wave emission timings, after
itation driven by intermediate focused ultrasound PNPs the fragmentation has occurred. The inset to Figure 9a is
of (a) 593 and (b) 863 kPa. At the lower amplitude, time-domain data captured around the moment of micro-
approximately f0-shock waves are apparent both before bubble cavitation contact with the capillary wall, at
and after the fragmentation event at »104 ms, at peak- »210 ms (Supplementary Video S5), indicating emis-
positive pressure values slightly higher than those emit- sions decrease to instrumental noise or below for the
ted during the 459-kPa driving (Fig. 4a). The shock remainder of the focused ultrasound burst.

Fig 8. Images extracted from a side-view shadowgraphic sequence at 10 Mfps, from t = 125 ms, of the microbubble cavi-
tation activity represented from top-view imaging in Figure 7, at peak negative pressures of (a) 593 kPa and (b) 863 kPa.
Images are background-subtracted to enhance shadowgraphic features, including the focused ultrasound driving and bub-
ble collapse shock waves, arrowed red. Bar = 150 mm; however, the cavitation activity appears larger, as the imaging is
purposefully defocused for shadowgraphic capability.
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2199

Fig. 9. (a, b) Needle hydrophone data, filtered and deconvolved from 2.4 to 20 MHz, to reveal shock wave content
within the microbubble cavitation emission signal, from the activity imaged in Figures 7 and 8 at peak negative pressures
of 593 and 863 kPa, respectively. In the inset to (a) are the data from around microbubble cavitation contact with the cap-
illary wall, at »210 ms. Black-dash boxes correspond to the duration of side-view shadowgraphic imaging, with shock
waves arrowed in Figure 8(a, b) similarly identified. (c) The spectra of (a) and (b), deconvolved over the full calibration
bandwidth, thereby include focused ultrasound driving. The needle hydrophone instrumental noise floor is also depicted
(orange dot). (d) Variance in shock wave emission times measured from (a) and (b). SW = shock wave.

At the higher-amplitude intermediate driving of 3 f0/2 = 2.175 ms). The spectra for both intermediate
PNP = 863 kPa, it can be seen that the microbubble cav- driving amplitudes (Fig. 9c) exhibit no clear peaks
itation switches regularly between f0- and f0/2-shock other than f0 and 2f0 and elevated noise floors relative
wave emission, also illustrated in Figure 9d to needle hydrophone instrumental noise, dispropor-
(green), with a mean period of 2.02 § 0.56 ms (with tionately around f0/2, for 863 kPa driving.
2200 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

Fig. 10. Representative images extracted from sequences of a multiple-microbubble observation, (a) top-view and (b)
side-view, at a peak-negative pressure of 459 kPa. (ce) Time domain data from the duration of side-view imaging, (f)
the spectrum (taken from t = 58175 ms) and (g) apparent shock wave emission interval, all presented relative to equiva-
lent data for single microbubble driven at 459 kPa (dotted black) of Figure 4c.

Multiple microbubble observation of driven microbubble cavitation emissions during therapy

Our primary aim was to collect data sets from single will involve detection from many microbubbles (which will
microbubble cavitation events, for ease of reconciling the also distributed through the vasculature), we present a sum-
acoustic emissions directly to the high-speed imaging, and mary of a multiple-microbubble data set in this section. The
shadowgraphic side-view capture of bubble collapse shock data were collected for focused ultrasound driving of
waves, in particular. During the course of repeating experi- PNP = 459 kPa and, so bear some comparison to the single-
ments to collect data from a relatively isolated microbubble microbubble cavitation at this PNP described previously.
aligned to the needle hydrophone, however, many multiple- Top-view imaging (Figure 10a; Supplementary
microbubble observations were also gathered. As monitoring Video S9, online only) indicates five microbubbles are
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2201

initially excited, with the emissions from the three PNPs, for microbubble cavitation emitting f0-shock
arrowed red and blue at t = 85.7 ms dominating the waves on collapse with every compressional phase. The
acoustic data sampled during side-view imaging lowest threshold PNP amplitude required to drive micro-
(Fig. 10cg), because of needle hydrophone directiv- bubble oscillation sufficiently for shock wave generation
ity. The two microbubbles arrowed red start cavitating has not been identified because of the limited spatial res-
at »44 mm from each other, appear to coalesce within olution of the imaging perspectives, preventing verifica-
30 ms and contact the capillary wall some 30 ms later. tion of any needle hydrophone data. We did, however,
Coincidentally, side-view image capture (Fig. 10b; observe f0-shock wave emission at a reduced PNP of 385
Supplementary Video S10, online only), occurs kPa (data not presented), suggesting that this response
around the moment of contact. Note, top-view imag- regime extends to amplitudes lower than that reported
ing reveals the cavitating bubble arrowed blue is above. At a PNP amplitude of 1.14 MPa, well in excess
»200 mm behind the coalesced bubble, along the axis of the fragmentation threshold, the resulting cavitation
of side-view imaging. Capillary wall contact is appar- cloud exhibits period-doubled oscillations, collapsing
ent as a cessation of translation, with the amplitude of sufficiently for shock wave emission only in response to
oscillation diminishing over several cycles of focused alternate compressional phases, raising spectral peaks at
ultrasound. Side-view imaging (Fig. 10b) at 132.1 ms nf0/2. At these higher driving PNPs, the amplitude of the
also suggests that the component cavitation bubbles bubble-cloud oscillation becomes sufficiently large that
may not have fully merged during the coalescence. the inertia of the host medium prevents full collapse,
Almost simultaneously, a cavitating bubble near the with every compressional phase. This behaviour has
top of the capillary, and therefore poorly focused in been reported previously, for single and apparently non-
top-view imaging (arrowed white in Fig. 10a), also fragmented microbubbles exposed to diagnostic imag-
contacts the capillary wall. Although only faintly ing-like frequencies (Chomas et al. 2002), but without
apparent, this microbubble cavitation gives some con- parallel acoustic detection. Indeed, such incomplete
fidence that driven cavitation-bubble activity at the oscillations were identified in early solutions to Ray-
top and bottom extremities of the capillary is percepti- leighPlesset-type equations, for single-bubble oscilla-
ble in top-view imaging, even at the lowest focused tion (Borotnikov and Solukhin [1964], as referred to in
ultrasound driving PNP presented. Neppiras (1980)), and was proposed as a candidate
The time-domain acoustic data in Figure 1c are laid mechanism, amongst others, for subharmonic generation
over the data of the single microbubble cavitation driven (Neppiras 1980).
at 459 kPa (from Fig. 4a), over the duration of side-view Microbubble cavitation behaviour that can contrib-
imaging, to facilitate comparison. Cursory inspection ute to an elevated noise floor in spectra, through broad-
suggests similarities in terms of f0-shock wave emission band emissions, is also investigated. The single,
times, but higher variance in peak-positive pressure un-fragmented, microbubble cavitation driven at a PNP
amplitude. Closer inspection (Fig. 10d, e), indicates of 459 kPa generated f0-shock waves with the lowest
unresolved multiple shock waves apparent as double variances, such that the noise floor of the spectrum,
peaks (d) or broadened full-width half maxima (e). The between the nf0 peaks, corresponded to instrumental
timings of component shock wave detection from the noise for the detector. For fragmented microbubble cavi-
individual cavitating microbubbles are similar, as the tation at higher PNPs, more pronounced variance in the
average shock wave propagation speed is similar to that shock wave emission interval timings and peak-positive
of the focused ultrasound driving each collapse. The pressure amplitudes effectively redistribute power from
spectrum of the detected emission signal features nf0 the spectral peaks, raising the floor to above instrumental
peaks, as for the single-microbubble cavitation driven by noise. Multiple-microbubble cavitation, driven at the
459 kPa, but with an elevated noise floor across frequen- lower PNP amplitude, generated shock waves from mul-
cies >2.5 MHz, by »10 dB. tiple sources that were incident to the needle hydrophone
tip at approximately the same time, such that similar nf0
peaks were raised in the spectrum. Differences in arrival
DISCUSSION
time of up to several hundred nanoseconds, for the com-
These results support periodic bubble collapse-gen- ponent shock waves from each source, however, were
erated shock waves as a mechanistic source of non-linear sufficient to raise the noise floor relative to the emission
acoustic emissions from microbubbles, driven below res- spectrum for the single-microbubble cavitation, across
onance and at parameters typical of those employed for higher frequencies. We also note that the imaged coales-
therapy. The period of shock wave emission determines cence event from the multiple-microbubble cavitation
the frequency values at which spectral peaks occur observations had no discernible effect on the regime of
(Song et al. 2016), with nf0 occurring for lower driving microbubble cavitation response or the eventual
2202 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

dissolution of the combined bubble on contact with the of the observations may also have relevance for the
capillary wall. mechanisms of tissue disruption. In particular, the con-
Blood vessel diameters in the brain range from »8 sistent observation that microbubble cavitation in f0-
mm for capillaries to several millimeters for the middle shock wave emission regimes did not survive contact
cerebral artery, with capillaries far outnumbering larger with the capillary wall could be significant. In contrast,
vessels. The 500-mm internal diameter of the polycar- the microbubble cavitation driven at 1.14 MPa, respond-
bonate capillary used for this work is therefore only ing in the f0/2 regime, endures after contact and contin-
comparable to a limited number of blood vessels in the ues to translate along the inner surface. If this is
brain, with smaller capillaries known to significantly indicative of microbubble cavitation in vasculature, then
suppress microbubble oscillation amplitude (Thomas the sustained mechanical activity in contact with the
et al. 2013; Caskey et al. 2006). Moreover, microbubble endothelial layer at higher PNP amplitudes could explain
cavitation in vivo will also be influenced by viscosity the more aggressive, often irreversible, bio-effects asso-
and the presence of blood cells (Sboros 2008). Further ciated with subharmonic emissions (O’Reilly and Hyny-
work is required to investigate the effect of reduced cap- nen 2012; Jones et al. 2018). Beyond physical contact,
illary diameter on periodic shock waves as the source of the role of bubble collapse shock waves for tissue disrup-
non-linear emissions from driven microbubbles. If, how- tion is a topic of ongoing research, with a thorough
ever, microbubble cavitation oscillations are suppressed, review recently available (Lopez-Marin et al. 2018). The
such that periodic shock waves are not generated as directivity of the 0.2-mm needle hydrophone used, and
described, raises the possibility that emissions detected its close proximity to the capillary, mean the peak-
during in vivo studies may result from a limited number positive pressure amplitudes of the shock wave measure-
of microbubble cavitation events, in larger blood vessels, ments reported are sensitive to the precise bubble loca-
and may not necessarily be representative of the micro- tion, relative to the needle tip, including as a result of
bubble population activity within the vasculature. translation of the activity during the focused ultrasound
To investigate single-microbubble cavitation behav- exposure. It is, however, notable that the shock waves
iour, and directly link the non-linear emissions generated emitted in the f0/2 regime are an order of magnitude
to that behaviour, we employed a highly dilute micro- greater than those emitted during f0-emission activity.
bubble solution, repeating experiments until single- Clearly, further work is required to verify these potential
microbubble cavitation was initiated within the target mechanisms in vivo.
region, defined by the location of the needle hydrophone In reports of applications of acoustic cavitation,
tip. Recent debate around safe bloodbrain barrier dis- generally, there is a tendency to classify activity as sta-
ruption suggests that lower concentrations may be ble or inertial, despite existing recognition that this cate-
favourable for avoidance of undesirable bio-effects, such gorisation is likely an oversimplification (Ashokkumar
as sterile inflammation (Kovacs et al. 2017a, 2017b; 2011; Leighton 1994), including for contrast agent
McMahon and Hynynen 2017, 2018; Silburt et al. microbubbles (Church and Carstensen 2001), with many
2017); however, most in vitro and in vivo experiments more subcategories required for a full description. In
will involve detection from many driven microbubbles. studies developing pre-clinical applications of focused
We have previously reported on the effects that a distrib- ultrasound and microbubbles for therapy in particular,
uted two-bubble cavitating system can have on the spec- stable cavitation is often associated with harmonic emis-
trum of the combined emissions (as recorded with a sions, broadband noise with inertial activity and subhar-
single-element detector), including spectral windowing monic emission as a signal indicative of transitioning
that can significantly suppress the key emissions associ- between the categories (the “upper limit” of stable cavi-
ated with the regime of bubble response (Song et al. tation). The work presented here further suggests that
2017). For microbubbles exposed to focused ultrasound this conventional categorisation is indeed inadequate
in the vasculature, where the resulting microbubble cavi- and, perhaps, unhelpful. Harmonic and subharmonic
tation activity will at least initially be spatially distrib- emissions are both mediated by collapse-generated
uted throughout the vessel network, spectral peak shock waves, a conventionally inertial effect. Moreover,
suppression in the emissions from neighbouring vessels elevated noise can result from variance in shock wave
could be an important and underrecognised effect, which emission timings and peak-positive amplitudes, occur-
could account for some of the variability reported for in ring after a fragmentation event under driving at inter-
vivo studies (Gorick et al. 2018). mediate PNP amplitudes, or from multiple-microbubble
Although the experiments described were intended cavitation driven by an f0-shock wave emission regime
primarily for identification of the mechanistic source of PNP.
non-linear emissions, during microbubble cavitation The results presented do not preclude the possibility
under subresonant focused ultrasound exposure, several of other mechanisms of non-linear emissions, such as
 Non-linear acoustic emissions from microbubbles
J. H. SONG et al. 2203

non-spherical oscillations, within a population of micro- and broadband emissions, can be accounted for through
bubbles exposed to focused ultrasound, for example, periodic shock wave characteristics, according to the pres-
some of which will be excited by lower pressure ampli- sure amplitude of the driving. Further research is required
tudes, outside or at the periphery of the focal region. The to validate these findings for in vivo microbubble cavita-
imaging FOVs for the current work, necessary to accom- tion for tissue disruption and drug delivery.
modate radiation force-induced translation and capturing
shock wave emission, have associated spatial resolutions Acknowledgments—The research leading to these results has received
funding from the European Research Council under the European
inadequate for investigating the interface of the bubble Union’s Seventh Framework Programme (FP/2007-2013)/ERC Grant
boundary. However, microbubble cavitation responding Agreement No. 336189 (TheraCav). We also acknowledge support
to an ultrasound exposure with the strong collapse from EPSRC network ThUNDDAR, in collaboration with Dr. James
McLaughlan at the University of Leeds, UK. The authors are very
phases necessary for periodic shock wave emission, grateful to Mr. Ewan Russell for preparing graphics associated with
including fragmentation events at higher amplitudes, is this work, Mr. Kristoffer Johansen for motivation and Professors Tony
unlikely to simultaneously sustain non-spherical oscilla- Gachagan and Sandy Cochran for support.
tion activity.
Finally, from all experiments performed, we note SUPPLEMENTARY MATERIALS
that we did not observe any variation in periodic shock
wave emission behaviour that might be attributable to Supplementary material associated with this article
the initial microbubble properties, such as quiescent can be found in the online version at doi:10.1016/j.ultra
equilibrium radius. As described under Methods, the res- smedbio.2019.04.005.
olution of both imaging perspectives is inadequate for
resolving microbubbles before focused ultrasound exci- REFERENCES
tation, however, more than 100 observations of micro-
Ashokkumar M. The characterization of acoustic cavitation bubbles—
bubble cavitation activity were taken at each of the PNP An overview. Ultrason Sonochem 2011;18:864–872.
amplitudes reported, the majority of which consist of Caskey CF, Kruse DE, Dayton PA, Kitano TK, Ferrara KW. Microbub-
multiple-microbubble cavitation events and single- ble oscillation in tubes with diameters of 12, 25, and 195 microns.
Appl Phys Lett 2006;88 033902.
microbubble cavitation events not sufficiently aligned to Caskey CF, Steiger Sm, Qin S, Dayton PA, Ferrara KW. Direct obser-
the needle hydrophone tip for synthetic signal recon- vations of ultrasound microbubble contrast agent interaction with
struction. Nonetheless, all observed microbubble cavita- the microvessel wall. J Acoust Soc Am 2007;122:1191–1200.
Chen H, Brayman AA, Kreider W, Bailey MR, Matula TJ. Observa-
tion displayed the general regime of periodic shock tions of translation and jetting of ultrasound-activated microbubbles
wave emission characteristics described above, in accor- in mesenteric microvessles. Ultrasound Med Biol 2011a;37:2139–
dance with the PNP of the focused ultrasound exposure. 2148.
Chen H, Kreider W, Brayman AA, Bailey MR, Matula TJ. Blood vessel
This is supported through comparison of the single- deformations on microsecond time scales by ultrasonic cavitation.
microbubble and multiple-microbubble cavitation obser- Phys Rev Lett 2011b;106 034301.
vations, both driven at a PNP of 459 kPa, for which all Chomas J, Dayton PA, May D, Ferrara KW. Nondestructive subhar-
monic imaging. IEEE Trans Ultason Ferroelectr Freq Control
component cavitation events collapse repeatedly at f0 2002;49:883–892.
and thereby emit f0-shock waves. Assuming that across Church CC, Carstensen EL. “Stable” inertial cavitation. Ultrasound
all observations for each PNP investigated, a representa- Med Biol 2001;27:1435–1437.
de Jong N, Emmer M, Chin CT, Bouakaz A, Mastik F, Lohse D, Ver-
tive cross section of the SonoVue microbubble subpopu- sluis M. “Compression-only” behavior of phospholipid-coated con-
lation, of initial diameter <5 mm, has been sampled, trast bubbles. Ultrasound Med Biol 2007;33:653–656.
suggests that the initial microbubble serves as a nucleus, Dollet B, van der Meer SM, Garbin V, de Jong N, Lohse D, Versluis M.
Nonspherical oscillations of ultrasound contrast agent-microbub-
with the acoustics dictating the subsequent cavitation bles. Ultrasound Med Biol 2008;34:1465–1473.
activity. This is in contrast to the strong dependence of Ferrara KW, Pollard R, Borden M. Ultrasound microbubble contrast
spherical subharmonic response in microbubbles excited agents: Fundamentals and application to gene and drug delivery.
Annual Rev Biomed Eng 2007;9:415–447.
at higher frequencies and lower driving amplitudes, for Gorick GM, Sheybani ND, Curley CT, Price RJ. Listening in on the
example, which was observed in »50 % of microbubbles microbubble crowd: Advanced acoustic monitoring for improved
and thought to depend critically on shell parameters (Sijl control of bloodbrain barrier opening with focused ultrasound.
Theranostics 2018;8:2988–2991.
et al. 2010). Johansen K, Song JH, Johnston K, Prentice P. Deconvolution of acous-
tically detected bubble-collapse shock waves. Ultrasonics
2017;73:144–153.
CONCLUSIONS Johnston K, Prentice P, Gerold B. Cavitation cloud translation in
focused ultrasound: Primary radiation force-mediated effects at
The role of periodic bubble collapse-generated shock varying intensity. Proc IEEE Int Ultrason Symp 2014;61–64.
waves within the acoustic emissions from therapeutically Jones RM, Deng LL, Leung K, McMahon D, O’Reilly MA, Hynynen
K. Three-dimensional transcranial microbubble imaging for guid-
driven contrast agent microbubbles is evaluated. All non- ing volumetric ultrasound-mediated bloodbrain barrier opening.
linear spectral features, including harmonic, subharmonic Theranostics 2018;8:2909–2926.
2204 Ultrasound in Medicine & Biology Volume 45, Number 8, 2019

Kovacs ZI, Kim S, Jikaria N, Milo B, Lewis BK, Bresler M, Burks SR, Sboros V. Response of contrast agents to ultrasound. Adv Drug Deliv
Frank JA. Disrupting the bloodbrain barrier by focused ultra- Rev 2008;60:1117–1136.
sound induces sterile inflammation. Proc Natl Acad Sci 2017a;114: Schneider M. Characteristics of SonoVueTM. Echocardiography
E75–E84. 1999;16:743–746.
Kovacs ZI, Burks SR, Frank JA. Concerning sterile inflammation fol- Sijl J, Overvelde M, Dollet B, Garbin V, Rozerndal T, de Jong N,
lowing focused ultrasound and microbubbles in the brain. Proc Natl Lohse D, Versluis M. Subharmonic behavior of phospholipid-
Acad Sci 2017b;114:E6737–E6738. coated ultrasound contrast agent microbubbles. J Acoust Soc Am
Kreider W, Crum LA, Bailey MR, Sapozhinkov OA. A reduced-order, 2010;128 32392352.
single-bubble cavitation model with applications to therapeutic Sijl J, Overvelde M, Dollet B, Garbin V, de Jong N, Lohse D, Versluis
ultrasound. J Acoust Soc Am 2011;130:3511–3530. M. “Compression-only” behavior: A second-order nonlinear
Kudo N. A simple technique for visualizing ultrasound fields without response of ultrasound contrast agent microbubbles. J Acoust Soc
Schlieren optics. Ultrasound Med Biol 2015;41:2071–2081. Am 2011;129:1729–1739.
Lajoinie G, Luan Y, Gelderblom E, Dollet B, Mastik F, Dewitte H, Silburt J, Lipsman N, Aubert I. Disrupting the blood-brain barrier with
Lentacker I, de Jong N, Versluis M. Non-spherical oscillations focused ultrasound: Perspectives on inflammation and regeneration.
drive the ultrasound-mediated release from targeted microbubbles. Proc Natl Acad Sci USA 2017;114:E6735–E6736.
Commun Phys 2018;1:22. Song J, Johansen K, Prentice P. An analysis of the acoustic cavitation
Leighton T. The acoustic bubble. London: Academic Press; 1994. noise spectrum: The role of periodic shock waves. J Acoust Soc
p. 413–524. Am 2016;140:2494–2505.
Lipsman N, Meng Y, Bethune AJ, Huang Y, Lam B, Masellis M, Song J, Johansen K, Prentice P. Covert cavitation: Spectral peak sup-
Herrman N, Heyn C, Aubert I, Boutet A, Smith GS, Hynynen pression in the acoustic emissions from spatially configured nuclea-
K, Black SE. Bloodbrain barrier opening in Alzheimer’s dis- tions. J Acoust Soc Am 2017;141:EL216–EL221.
ease using MR-guided focused ultrasound. Nat Commun 2018; Sun T, Zhnag YZ, Power C, Alexander PM, Sutton JT, Aryal M,
9:2336. Vykhodtseva N, Miller EL, McDannold NJ. Closed-loop control of
Lopez-Marin LM, Rivera AL, Fernandez F, Loske AM. Shock wave- targeted ultrasound drug delivery across the bloodbrain/tumor
induced permeabilization of mammalian cells. Phys Life Rev barriers in a rat glioma model. Proc Natl Acad of Sci USA
2018;26/27:1–38. 2017;114:E10281–E10290.
McDannold N, Vykhodtseva N, Hynynen K. Targeted disruption of the Thomas DH, Sboros V, Emmer M, Vos H, de Jong N. Microbubble
bloodbrain barrier with focused ultrasound: Association with oscillations in capillary tubes. IEEE Trans Ultason Ferroelectr Freq
cavitation activity. Phys Med Biol 2006;51:793–807. Control 2013;60:105–114.
McMahon D, Hynynen K. Acute inflammatory response following van der Meer SM, Versluis M, Lohse D, Chin CT, Bouakaz A, de Jong
increased bloodbrain barrier permeability induced by focused N. The resonance frequency of SonoVueTM. Proc IEEE Int Ultrason
ultrasound is dependent on microbubble dose. Theranostics Symp 2004;343–345.
2017;7:3989–4000. van Wamel A, Kooiman K, Harteveld M, Emmer M, ten Cate FJ, Ver-
McMahon D, Hynynen K. Concerning acute inflammatory response sluis M, de Jong N. Vibrating microbubbles poking individual cells:
following focused ultrasound and microbubbles in the brain. Thera- Drug transfer into cells via sonoporation. J Controlled Release
nostics 2018;8:2249–2250. 2006;112:149–155.
Neppiras E. Acoustic cavitation. Phys Rep 1980;61:159–251. Versluis M, Goertz DE, Palanchon P, Heitman IL, van der Meer SM, Dol-
O’Reilly MA, Hynynen K. Bloodbrain barrier: Real-time feedback- let B, de Jong N, Lohse D. Microbubble shape oscillations excited
controlled focused ultrasound disruption by using an acoustic through ultrasonic parametric driving. Phys Rev E 2010;82(2) 026321.
emissions-based controller. Radiology 2012;263:96–106. Wu SY, Tung YS, Marquet F, Downs ME, Sanchez CS, Chen CC, Ferrara
Prentice P, Cuschieri A, Dholakia K, Prausnitz M, Campbell P. Mem- V, Konofagou E. Transcranial cavitation detection in primates during
brane disruption by optically controlled microbubble cavitation. bloodbrain barrier opening—A performance assessment study.
Nat Phys 2005;1:107–110. IEEE Trans Ultason Ferroelectr Freq Control 2014;61:966–978.