Nanobubble Contrast Enhanced Ultrasound Imaging: A Review

ABSTRACT

Contrast‐enhanced ultrasound is currently used worldwide with clinical indications in cardiology and radiology, and it continues to evolve and develop through innovative technological advancements. Clinically utilized contrast agents for ultrasound consist of hydrophobic gas microbubbles stabilized with a biocompatible shell. These agents are used commonly in echocardiography, with emerging applications in cancer diagnosis and therapy. Microbubbles are a blood pool agent with diameters between 1 and 10 μm, which precludes their use in other extravascular applications. To expand the potential use of contrast‐enhanced ultrasound beyond intravascular applications, sub‐micron agents, often called nanobubbles or ultra‐fine bubbles, have recently emerged as a promising tool. Combining the principles of ultrasound imaging with the unique properties of nanobubbles (high concentration and small size), recent work has established their imaging potential. Contrast‐enhanced ultrasound imaging using these agents continues to gain traction, with new studies establishing novel imaging applications. We highlight the recent achievements in nonlinear nanobubble contrast imaging, including a discussion on nanobubble formulations and their acoustic characteristics. Ultrasound imaging with nanobubbles is still in its early stages, but it has shown great potential in preclinical research and animal studies. We highlight unexplored areas of research where the capabilities of nanobubbles may offer new advantages. As technology advances, this technique may find applications in various areas of medicine, including cancer detection and treatment, cardiovascular imaging, and drug delivery.

(A) Numerous nanobubble (NB; ~100–500 nm diameter) formulations have been studied for imaging applications. Most agents are (i) stabilized with a protein, polymer, or lipid shell. Several formulations have also been (ii) decorated with fluorescent labeling or (iii) functionalized with epitopes targeting overexpressed biomarkers on diseased tissue. (B) NBs have been tested across a broad range of ultrasound (US) frequencies for imaging purposes. They produce contrast because of the compressibility and density of the gas cavity and the nonlinear response to the incident wave controlled by the shell characteristics. (C) NBs are injected into the bloodstream. (i) In healthy, non‐leaky vasculature, NBs are predominantly intravascular agents. (ii) NBs have been shown to interact with red blood cells in a “hitchhiking” fashion, which may extend their circulation time. (iii) NBs can extravasate in the presence of leaky vasculature. (iv) Following extravasation, NBs can passively accumulate in the extracellular matrix of tumors, (v) actively bind to targeted cells, and (vi) internalize in a cell. Made with Biorender.

1. Introduction

The application of ultrasound (US) to medical imaging dates back more than 65 years and has evolved into a standard medical imaging system, hailed for its excellent safety record, and used in nearly all hospitals and clinics worldwide (Food and Drug Administration ²⁰²³; Jensen ²⁰⁰⁷). The key strengths of diagnostic US are its ability to reveal the anatomy, the dynamic movement of organs, and real‐time blood flow details (Szabo 2014). The US wave, generated by a transducer (typically in the 1–30 MHz range for most medical imaging applications), is either reflected, scattered, or absorbed by tissue and boundaries between tissues, such as skin and bone. In imaging, the timing and strength of the returned echoes are used to determine the location of boundaries and/or scatterers in the field of view.

US contrast agents were developed after it was discovered that increased reflectivity was caused by gas microbubbles (MBs) (Szabo 2014). Surfactant‐stabilized MBs (1–10 μm diameter) are now the dominant agents used for US enhancement (see Figure 1; Sirsi and Borden ²⁰⁰⁹; W. Zeng, Yue, and Dai ²⁰²³). Currently, there are a number of commercially available MBs that are used in clinics (see Table 1). All clinically utilized US contrast agents consist of a shell‐encapsulated heavy gas core. The agents reported in Table 1 are marketed as microspheres ranging from ~1 to 5 μm in diameter. MBs have United States Food and Drug Administration approval for the enhancement of echocardiographs and detection of liver lesions (Frinking et al. 2020) in adult and pediatric populations (Sidhu et al. 2017) and have been investigated widely in numerous other clinical diagnostic applications. Several excellent reviews of MBs in US imaging and related applications have been previously published (Fournier, de La Taille, and Chauvierre ²⁰²³; Frinking et al. ²⁰²⁰; Upadhyay and Dalvi ²⁰¹⁹).

FIGURE 1.

Diagram comparing the dimensions of a typical microbubble, to a nanobubble and a typical gas vesicle. Made with Biorender.

TABLE 1.

Summary of clinically used, commercially available bubble formulations for contrast‐enhanced ultrasound.

Contrast agent	Shell material	Gas core	Marketed diameter	Concentration	References
Definity	Phospholipid	Perflutren	1.1–3.3 μm; 98% < 10 μm	1.2 × 1010 mL−1	(DEFINITY® Prescribing Information 2024)
Lumason (SonoVue)	Phospholipid	Sulfur hexafluoride	2.5 μm; 90% < 8 μm	100–500 × 106 mL−1	(Schneider 1999)
Optison	Human albumin	Perflutren	3–4.5 μm (max. 32 μm); 95% < 10 μm	5–8 × 108 mL−1	(Optison™ 2021)
Sonazoid	Hydrogenated egg phosphatidyl serine	Perflubutane	2.6 μm volume‐median diameter	1.2 × 109 mL−1 0.1% > 7 μm	(Sontum 2008)

While the composition of MBs is varied, the mechanisms underlying their ability to generate strong US image enhancement are similar. When placed in an acoustic field, bubbles undergo volumetric oscillation in response to positive and negative pressure changes in a manner distinct from the surrounding medium due to differences in the compressibility and density between the gas and blood and mediated by the viscoelastic properties of the shell. However, MB applications are limited by a relatively large particle diameter (~1–10 μm), which confines the MB to the vasculature. As such, information provided by MBs is limited to disease biomarkers within the intravascular space.

Nanobubbles (NBs) for US are an emerging class of sub‐micron contrast agents (i.e., Figure 1). The main advantage of using NBs for US imaging is the novel ability to combine the extravascular access of nano‐scale agents with the exceptional safety profile and low cost of US. NBs can be used in new imaging applications due to their ability to flow through microcapillaries in relatively high concentration and to move beyond the vascular endothelium when leaky vasculature is present (e.g., inflammation). Herein, the NBs to which we refer are bulk NBs (freely suspended spherical bubbles) and are approximately an order of magnitude smaller in diameter compared to MBs. The history of NBs is a tortuous one. The existence of NBs was long‐disputed due to estimates of large increases in surface tension as a function of the reduced bubble radius (Alheshibri et al. 2016; Kyzas and Mitropoulos ²⁰²¹; X. Wang, Li, et al. ²⁰²³). Today, the source of the stability of bulk NBs is better understood (Kyzas and Mitropoulos 2021; Nirmalkar, Pacek, and Barigou ^2018b; Zhou et al. ²⁰²¹). While there are NB structural and stability differences for different applications, NB research is now applied to drinking water (Atkinson et al. 2019; Singh et al. ²⁰²¹), wastewater treatment (Temesgen and Han 2023), surface cleaning (N. Jin, Zhang, et al. ²⁰²²), medical imaging (R. Wu et al. ²⁰²¹; Xu ²⁰¹¹), and nutritional applications such as fishery and agriculture (Ebina et al. 2013). Numerous reviews have described NB production methods, properties, and potential applications (Favvas et al. 2021; Foudas et al. ²⁰²³; Hansen et al. ²⁰²³; Paknahad et al. ²⁰²¹; Zhou et al. ²⁰²¹). The reader is encouraged to read the review article (Alheshibri et al. 2016) for a detailed account of NB history.

NB use in US imaging has significantly increased since the first US‐oriented application was reported in 2004 (Oeffinger and Wheatley 2004). Numerous publications have demonstrated strong and prolonged nonlinear signal in vitro and in vivo. Since their size allows situational NB transport to target sites beyond the vasculature, NBs have been of particular interest in molecular US imaging (F. Zeng, Du, and Chen ²⁰²¹; Zhang et al. ²⁰²²) and as a means of determining the tissue penetration of nanotherapeutics. The utilization of NBs in US has witnessed a significant rise in recent years, meriting review.

We present an overview of NBs in contrast‐enhanced US (CEUS). We first provide a summary of the reported NB formulations and methods designed for CEUS using NBs (NB‐CEUS). Next, we explore the fundamental physics of NB characteristics that are critical for effective NB‐CEUS and draw on principles first explored for MB use. Pulse sequences designed to generate bubble‐based contrast are discussed, and progress toward NB‐specific pulse sequences is highlighted. Next, we discuss exciting advancements in the applications of CEUS using NBs, novel imaging and image analysis techniques, and areas of research with opportunities for further investigation.

2. Sub‐Micron Contrast Agent Formulation and Characterization

Here, we present an overview of previously reported NB formulations for contrast‐enhanced diagnostic imaging and therapeutic applications. Our focus is gas‐core shell‐stabilized NBs, closely related to the commercial MB formulations. We also discuss sub‐micron components of commercial MB formulations. Finally, we include a brief discussion of gas vesicles (GVs), a unique set of submicron gas‐core agents, distinct from NBs, that are being employed and are relevant to NB research owing to their similarity in size and scattering characteristics.

It is important to note that the list of sub‐micron US contrast agents provided here is non‐exhaustive. For instance, phase change nanodroplets are liquid perfluorocarbon nanospheres with low boiling temperatures which are forced into a gaseous microsphere by the application of an external energy source (e.g., laser irradiation or focused US) (Aliabouzar et al. 2023; Durham and Dayton ²⁰²¹; Strohm et al. ²⁰¹¹; Kakaei et al. ²⁰²³; Zhang et al. ²⁰²³). However, these agents are typically only nanoscale (150–250 nm) during their liquid phase before vaporization. Once vaporized into gaseous form, these agents can span on average 5–10 times the initial droplet diameter (vaporized diameter 1.5–2.5 μm) (Pellow et al. 2024). The advantages of nanodroplets are the prolonged circulation over MBs, increased ability for payload delivery in therapeutic applications, and opportunity for improved localization of cavitation (Lea‐Banks, O'Reilly, and Hynynen ²⁰¹⁹; Shakya et al. ²⁰²⁴). Another example of sub‐micron agents previously explored for US‐mediated therapy is liposomes, not outlined here.

2.1. Nanobubbles

The NB formulations used for US imaging typically involve a stabilizing protein, polymer, or lipid shell which encases a heavy gas core (usually a perfluorocarbon). Some formulations are bio‐inspired, lipid‐based structures with a shell produced by a lipid mixture with stabilizers (i.e., cholesterol) not unlike the cell wall. Importantly, the choice of materials employed for the agent production plays an important role in the contrast generation in CEUS, and, therefore, must be carefully considered. We provide a summary of NB formulations proposed for US imaging applications and their respective production methods and characteristics in Table 1. The production methods for biomedical NBs involve several commonalities: (1) a stabilizing shell, (2) a low solubility core, and (3) high agent yield (> 10¹⁰).

The shell material is usually chosen based on stability, biocompatibility, cost, and functionalization requirements and ultimately depends on the final application. Phospholipids consist of a hydrophilic head and hydrophobic tail, which, under appropriate conditions, can self‐assemble into a single layer in an aqueous environment to form NBs around the gas core. The thin (~5 nm), flexible lipid shell (usually a mixture of several lipids (Kumar et al. 2022; X. Wang, Liu, et al. ²⁰²³; L. Zhu et al. ²⁰¹⁷)) is held together by weak intermolecular forces. These weak forces allow for the expansion and contraction of the NB in the presence of pressure waves (Upadhyay and Dalvi 2019). The flexibility of lipid‐shelled NBs contributes to nonlinear volumetric oscillations and increased CEUS signal, which is likely why most NB formulations employ a phospholipid shell. However, the weak intermolecular forces holding the lipid shell together result in a lower shell density, allowing easier gas diffusion from the core of the bubble and decreasing the stability (Upadhyay and Dalvi 2019). NB formulations which are stabilized by a lipid shell tend to have high lipid concentrations (ranging between 2 and 14 mg/mL) compared to commercial MBs (Definity: 0.75 mg/mL; Lumason: 5 mg/mL). High lipid concentrations are particularly important for NB formulations since low lipid concentrations lead to bubble coalescence, as previously demonstrated in microfluidic production of MBs (Segers et al. 2016). With high lipid concentrations, mechanical compression of the lipid mono‐layer results in a surface pressure high enough to counteract, or balance, the surface tension while decreasing gas diffusivity through the membrane. The high lipid concentration also results in the presence of lipid structures such as micelles and liposomes in the bubble solution that act to minimize bubble coalescence (Batchelor et al. 2022).

The viscoelastic properties of the lipid shell can be altered by introducing membrane additives, such as propylene glycol as a membrane softener or glycerol as a membrane stiffener (Jafari Sojahrood et al. 2021). Alternatively, polymers (e.g., poly(lactic‐co‐glycolic acid); Du et al. ²⁰¹⁸) can be used to produce a NB with a ‘hard shell’ due to dense packing (Rudakovskaya et al. 2022). A dense shell offers an increased ability to store drugs and resist gas diffusion but also has a trade‐off of reduced contrast generated under US due to increased resistance to volumetric oscillations. As a result, compared to lipid shells, polymer‐shelled NBs could require higher acoustic pressures to obtain adequate US contrast (P. V. Chitnis et al. ²⁰¹³). Meanwhile, protein shells (e.g., chitosan polyacrylic acid (Yi et al. 2023)) are held together with strong noncovalent intermolecular forces, which can be further strengthened via cross‐linking (Rovers et al. 2016). The strength of the intermolecular forces and degree of cross‐linking play a role in determining the acoustic pressure required to generate adequate contrast signal. Additionally, protein‐shelled NBs may activate an immunological effect upon administration (Upadhyay and Dalvi 2019). When determining the type of NB shell required for an application, it is critical to consider the NB stability, bio‐compatibility, and US image acquisition parameters.

The compressible gas core of NBs is critical for the volumetric oscillations, which contribute to strong reflection of US and therefore high signal. Heavy gases such as perfluorocarbons are commonly used in the core due to their low solubility in water, which increases their stability (Schutt et al. 2003). The Optison formulation displayed improved stability over Albunex because it employed octofluoropropane as the gas core instead of air despite using the same encapsulating shell (Podell et al. 1999). The lower solubility of perfluorocarbon delays the escape of encapsulated gas from the agent. While gas properties have predominantly been explored for MB formulations, this same principle applies to NB formulations.

Production methods of NB formulations are numerous. As shown in Table 1, the most common method of NB production involves the activation of a highly polydisperse population followed by the isolation of a size‐selected population subset. This could involve mechanical agitation followed by differential centrifugation (de Leon et al. ²⁰¹⁹; L. Duan et al. ²⁰²⁰; Pellow et al. ²⁰¹⁸; R. Perera et al. ²⁰¹⁹; L. Zhu et al. ²⁰¹⁷). Alternatively, another common technique involves the reconstitution of a dehydrated film (W. B. Cai et al. ²⁰¹⁵; X. Wang, Liu, et al. ²⁰²³). Such techniques (mechanical agitation‐based or rehydration‐based) offer the advantage of high‐yield agent production. However, the filtration processes involved in the isolation of the population subset (i.e., differential centrifugation or pore filtration) generate waste and contribute to agent destabilization by shear stress. More recent approaches have attempted to produce the NB formulations while minimizing waste using microfluidics (Paknahad et al. 2021). For instance, a technique proposed by Kumar et al., involved the individual shrinking of MBs into NBs while funneling agents though a microfluidic network (Kumar et al. 2022). Recent advances in microfluidic techniques enabled high‐concentration NB production (Peyman et al. 2016). Another novel technique proposed by Counil et al. involved the use of an extrusion process (commonly used for liposomal production) to produce zero‐waste high‐volume yield native NB (Counil et al. 2022). Most noteworthy here is the rapid and efficient high yield of NB monodisperse populations with easily tunable size. Since agent polydispersity plays an important role in image quality, production methods with monodisperse outputs will likely grow in popularity.

The method to employ for agent production must be carefully considered. For in vivo applications, a highly concentrated sample with a high yield for bolus injection is often desirable. Common injections have a final diluted concentration of ~10¹⁰ NBs/mL in the body to generate a sufficiently high signal (a bolus of 50–200 μL from ~10¹¹ NBs/mL stock is common (Abenojar et al. 2019)). For agents that are being used for theranostic applications, it might be desirable to have a less polydisperse population since monodisperse agents have a more sensitive response to US parameters than polydisperse populations. For proof‐of‐concept studies, it might be more advisable to use a microfluidics technique because it can produce a monodisperse population with homogeneous shell properties (P. V. Chitnis et al. ²⁰¹³). In sum, NB formulation and production is an extensive topic which has merited significant attention (Favvas et al. 2021; Zhou et al. ²⁰²¹). A thorough assessment of the production methods merits its own review and is beyond the scope of this paper.

2.2. Gas Vesicles

Gas Vesicles (GVs) are a unique class of air‐filled protein nanostructures naturally produced in buoyant microbes (Lakshmanan et al. 2017; Ling et al. ²⁰²⁴). Microbes such as the cyanobacterium Anabaena flos‐aquae and the archaeon Halobacteria salinarum form GVs as a means to regulate cellular buoyancy. GVs can be purified from cyanobacteria and haloarchaea and used to produce US contrast (Lakshmanan et al. 2017). They are comprised of a corrugated protein shell of varying thickness (1–3 nm) that excludes liquid water while allowing dynamic gas exchange in and out of their interior (Shapiro et al. 2014). GVs can be produced in highly monodisperse populations by the micro‐organisms. They are frequently cylindrical or spindle‐shaped protein encapsulations with 100 nm–2 μm in length and 45–200 nm in width (Figure 1; W. Zeng, Yue, and Dai ²⁰²³).

While they share numerous similarities with NBs, it is important to recognize that GVs are distinct from NBs and should not be termed interchangeably. While their small scale allows them to extravasate beyond the vascular network in leaky vasculature, structural differences which distinguish them from NBs play an important role in the US response of the agent. Therefore, the optimal US exposure parameters for NBs are unlikely to be optimal for GVs of the same size.

Acoustic reporter genes that produce GVs have been shown to act as effective markers to image and locate microorganisms in mammalian hosts (Schnell 2018). As research on the physiological interactions of GVs progresses, such as recent investigations on the interactions of GVs with blood components (Ling, Ko, et al. ²⁰²³), our understanding of gas‐core agent bioaccumulation improves. Owing to the similar size scale and therefore overlap in applications, research on GVs will likely translate to improvements in the field of NBs. One example is the nonlinear X‐wave US imaging pulse sequences used with GVs (e.g., cross‐amplitude modulation (xAM)) to minimize cumulative nonlinear US propagation effects that produce image artifacts from linear scatterers (Maresca et al. 2018). There is room for the translation of these methods to NB‐CEUS. Research on GVs is extensive. Several groups have published reviews on GVs, to which the interested reader is directed for more information (Pfeifer 2012; Walsby ¹⁹⁹⁴; R. Wang et al. ²⁰²²; W. Zeng, Yue, and Dai ²⁰²³).

2.3. Submicron Components of Commercially Available Bubble Agents

Only one commercially available NB (150–200 nm; 1.8 × 10¹² mL⁻¹) for US has been identified at this time, and this agent is for research use only (Ultrasound Nanobubbles (NB‐20), FUS Instruments). Otherwise, most commercially available agents are marketed as MBs or microspheres. However, it is worth noting that these agents exhibit a high degree of polydispersity and have significant nanoscale populations. As such, the marketed diameters reported in Table 1 does not reflect the actual size range of the agents. For instance, over a decade ago, Goertz et al. reported on the presence of submicron populations in Definity (marketed as 1.1–3.3 μm diameter) (Goertz, de Jong, and van der Steen ²⁰⁰⁷). A prevalent nanoscale population was confirmed even when increasing to physiologically relevant temperatures (Shekhar et al. 2018) and helps explain the unusually high attenuation coefficient by Definity at as high as 25 MHz—well beyond the resonance peak (~2–7 MHz) (Raymond et al. 2013). Similar to Definity, number‐weighted distributions from Coulter‐based measurement instruments indicate significant sub‐micron populations in many of the currently available commercial formulations (Figure 2). However, since the lower limit of detection for these Coulter‐based measurements is often as high as ~600–700 nm, it is difficult to appreciate the actual particle range. More details about bubble characterization methods are discussed below.

FIGURE 2.

Coulter counter (Multisizer 4, Beckman Coulter Brea, CA, USA; 30‐μm aperture tube, 100‐μL volumetric setting; 0.6–18 μm) size distribution measurements for (a) Definity, (b) MicroMarker. Mean number density (histogram, shaded) and volume‐weighted size distribution (solid line) are plotted versus particle diameter. Reproduced from Raymond et al. (2013). (c) Number‐weighted size distribution measurements (Multisizer 3, 30‐μm aperture tube, measurement range 0.7–18 μm, Beckman Coulter, Fullerton, CA, USA) of various commercially available contrast agents (Lumason‐blue, Definity‐yellow; Optison‐green) and two in‐house prepared agents before (BG8758‐black) and after (MSB4) size isolation. Reproduced from Helbert et al. (2020) with permissions requested.

2.4. Characterization Techniques for Submicron Agents

Characterizing the size and concentration of bubble populations is quite complex due to their polydispersity, buoyancy, deformability, and fragility. This is especially true for commercially available MB formulations like in Table 1, which owe their significant polydispersity to variable ‘activation’ or agitation protocols. Few, if any, technologies offer the capability of measuring size and concentration of particles across all scales of interest (~50 nm—20 μm) in a single acquisition. Thus, a suite of characterization methods has been proposed to confirm the size distribution and concentration of these compressible agents (Exner and Kolios 2021). Several techniques have been employed to characterize mono‐ and polydisperse bubble populations, including resonant mass measurement (RMM), dynamic light scattering (DLS), nanoparticle tracking analysis (NTA), gas chromatography/mass spectroscopy (GC/MS), and Coulter‐based sizing instruments (such as the Multisizer, from Beckman Coulter). Each technique offers unique strengths and weaknesses.

The RMM technology utilizes a microfluidic channel embedded in a resonating cantilever under vacuum to detect, count, and measure the buoyant mass of particles in a liquid passing through the channel. Shifts in the resonant frequency of the cantilever occur as particles flow through the microfluidic channel. The direction of the frequency shift is used to classify the particle as either buoyant or non‐buoyant, while the amplitude of the shift is used to determine the particle diameter. The application of RMM to NB sizing and MB sizing has been studied previously (Hernandez et al. 2019). The unique advantage of RMM‐based characterization is the ability to distinguish between buoyant and non‐buoyant particles. However, the availability of these instruments is currently limited.

DLS uses the principle of a size‐dependent diffusion coefficient of particles in suspension to produce a measure of hydrodynamic radii. A historical timeline and overview of DLS principles can be found in (Stetefeld, McKenna, and Patel ²⁰¹⁶). In the review by Stetefeld et al., several limitations of the DLS technique are listed. The size‐dependent error is particularly important for bubble populations, which increases for larger bubbles as buoyant force increases. For instance, the rise velocity is ~5.2 × 10⁻⁴ mm/s for a 1 μm MB and ~5.2 × 10⁻⁶ mm/s for a 100 nm NB (Goertz, de Jong, and van der Steen ²⁰⁰⁷). However, from the Stokes–Einstein equation for translational diffusion, the mean Brownian velocity at 25°C for 100 nm, and 1 μm diameter particles are 3 × 10⁻³ mm/s, and 0.9 × 10⁻³ mm/s respectively. The terminal velocity caused by buoyant force thus results in ~18% increase in velocity for a 1 μm MB but only 0.6% increase in velocity for a 100 nm NB for an air bubble in water. However, the pressurized heavy gas core (density = 8 kg/m³) combined with the added mass of the surfactant shell reduces the buoyant force. Previous reports have also shown that, for NBs, the DLS technique overestimates diameter compared to other techniques (Ma et al. 2022). Additionally, DLS machines do not usually provide measures of concentration. The unique advantage of DLS‐based sizing is the large range of particle sizes (from 0.3 nm to > 10 μm (e.g., Nanotrak Flex, Microtrac Retsch GmbH)) which can be measured in a single acquisition, providing the most appropriate measure of polydispersity for a bubble population. DLS is currently the prime experimental technique for detecting bulk NBs (Kyzas and Mitropoulos 2021). More recently, NTA was developed for tracking and sizing bulk NBs and Brownian diffusing particles. The NTA technique is related to DLS in that it is an optical method. Scattered light spots reflect the particle position, and the Brownian motion of each particle in the field of view is recorded, enabling “tracking” measurements. NTA sacrifices sizing range to produce a superior resolution for nanoparticles and measures of concentration (Ma, Li, and Sun ²⁰²²).

To quantify the amount of gas within the NBs, headspace GC/MS can be used. GC/MS is considered the gold standard for gas volume quantification and is used to assess the accuracy of other techniques. For instance, in comparing RMM to Coulter, GC/MS was used to identify the overestimation of gas volume by Coulter‐based methods (Abenojar et al. 2019; JafariSojahrood et al. ²⁰¹⁷). NBs can also be visualized with microscopy, including transmission electron, scanning electron, cryo‐electron, optical, and fluorescent microscopy (Abenojar et al. 2019; Counil et al. ²⁰²²; Hernandez et al. ²⁰¹⁷; Wegierak et al. ²⁰²²). However, light microscopy can be a challenge because the average diameter NB is lower than the diffraction limit. Meanwhile, low throughput and limited visibility potentially make size measurements nonrepresentative in electron microscopy techniques.

Coulter‐based measurements of bubble diameter have been the standard in the field to characterize commercially available bubble agents (Sennoga et al. 2012). However, these techniques require pressurized flow focusing through an aperture, which is known to reshape deformable particles, including cells (Taraconat et al. 2019). Additionally, reports of average diameter often require data re‐weighting to obtain a peak by plotting volume‐weighted distributions instead of number‐weighted distributions (see Figure 2). While the volume of particles is important for US contrast agents owing to the increased scattering cross‐section of larger agents, this re‐weighting biases size measurements toward larger diameters and imposes artificial peaks in the distribution while underestimating the contribution of agents closer to the minimum detection limit (aperture‐tube dependent). For example, for most models including the currently commercially available Beckman Multisizer TM 4 equipped with a 20 μm aperture there is a lower limit of detection of ≥ 400 nm. Meanwhile, the most common protocol for commercial‐agent sizing is the 30 μm aperture which has a lower limit of detection of 600 nm (see Figure 2). Additionally, problems with the coincidence of small particles traversing the Coulter Counter aperture can be a significant hurdle causing false recordings of large particles when two smaller ones traverse the aperture (Wynn and Hounslow 1997). Coincidence issues can be partially addressed by sample dilution or signal processing corrections. However, they continue to be a known limitation (Caselli et al. 2021).

Overall, any of the techniques mentioned above have advantages and disadvantages that further complicate the characterization of the agent populations. Only when multiple techniques are used can mono‐ or polydisperse NB populations be confirmed. While MB sizing methods have been previously reviewed (Sennoga et al. 2012), an updated review is needed to summarize the growth and development in the measurement of size distributions that span all scales of interest (~50 nm—20 μm) in a single acquisition.

3. Nanobubble US Response

The contrast generated by NBs in CEUS primarily relies on the contrast agents' nonlinear and tissues' linear responses to an US wave. The bubbles' response to the incident wave is highly sensitive to initial conditions, including bubble size, shell properties, and the surrounding environment. If the characteristics of the agent and the environment remain fixed, the pulse characteristics (pressure, frequency, and duration) of the US wave determine the bubble behavior. There are many excellent reviews of MB CEUS and the physics involved to which the reader is directed for comprehensive knowledge of bubble dynamics (Calliada et al. 1998; De Jong, Bouakaz, and Frinking ²⁰⁰²; Goldberg, Liu, and Forsberg ¹⁹⁹⁴; Sirsi and Borden ²⁰⁰⁹; Upadhyay and Dalvi ²⁰¹⁹). We first describe fundamental concepts in CEUS before reviewing NB‐specific studies.

3.1. Bubble‐US Interactions

As US waves encounter a deformable gas‐filled particle, it is subject to the compression and rarefaction of the pressure wave. In a simple low‐pressure case (typically < 50 kPa), the bubble oscillates linearly and, like tissue, produces backscatter that is proportional to the driving pressure (De Jong, Bouakaz, and Frinking ²⁰⁰²). For moderate acoustic pressures (50–200 kPa), nonlinear oscillations occur, giving rise to harmonics (including ultra‐ and super‐harmonics) and subharmonics. Nonlinear response can also occur close to bubble resonance—a bubble will demonstrate prolonged increased oscillations when sonicated at its resonance frequency (3.8 MHz for 2 μm diameter MB (De Jong, Bouakaz, and Frinking ²⁰⁰²)). The nonlinear behavior of MBs has been thought to be the main driver for effective contrast enhancement in CEUS. Thus, numerous studies have explored the parameters necessary for the generation of periodicities, high‐order harmonics and chaos (Sojahrood et al. 2019; Sojahrood, Haghi, Karshafian, et al. ^2021b; Sojahrood, Haghi, Porter, et al. ²⁰²¹). Harmonic nonlinear signals can be isolated by frequency filtering or by using imaging pulse sequences with a modulated phase, frequency, and amplitude to separate agent signals from background tissue signals, further enhancing image contrast (Sirsi and Borden 2009).

3.2. US Pulse Sequences

US imaging employing contrast agents is rarely, if ever, conducted in fundamental mode due to high tissue backscatter at the fundamental frequency causing reduced image contrast. It is the isolation of the nonlinear components of bubble backscatter from US which enables CEUS. Image acquisition techniques use pulse sequences designed to amplify the signal from nonlinear scatterers to further enhance and isolate the contrast signals from MBs. Typically, the goal of these sequences is to suppress the linear portion of the signal from the tissue, thereby increasing the relative contribution of nonlinear signals from the US contrast agents injected into the bloodstream. The two most common are pulse inversion (PI) and pulse amplitude modulation (PAM). Both techniques are dependent on the nonlinear dynamics of ultrasonically excited agents.

The PI technique involves two pulses with equal amplitude and frequency but inverted waveforms (Shen, Chou, and Li ²⁰⁰⁵). When the returning echoes from the two pulses are summed, the residual signal from tissue is limited, while the residue signal from MBs is not entirely canceled because the volumetric change of the bubbles under compression is different from that under rarefaction (Morgan, Averkiou, and Ferrara ¹⁹⁹⁸). Morgan et al. found high echo amplitudes coincided with rarefactional half‐cycles, where the mean frequency of the agent echo depended on the transmitted phase. When the acoustic wave leads with rarefaction, the mean echo frequency shifts upward (Morgan, Averkiou, and Ferrara ¹⁹⁹⁸). Thus, the bubble echoes from inverted pulses will not cancel when summed. The backscatter from tissue behaves differently. Since sound propagates nonlinearly at higher pressures (Harpen 2006; Humphrey ²⁰⁰⁰), the backscattered echo from tissue can have harmonic components, which can be modeled as a power series (Shen and Li 2003). When summing the tissue backscatter of two inverted pulses, the sum leaves only even‐ordered residuals. Since bubble agent behavior cannot be modeled by the power series, tissue signal is canceled more than bubble echo (Shen, Chou, and Li ²⁰⁰⁵). PI sequences have since been implemented to retain the second‐order nonlinear term of echo backscatter from bubble agents.

By comparison, PAM takes advantage of pressure‐dependent nonlinear bubble behavior. In the simplest case, two pulses are fired, one with high enough pressure to drive a nonlinear bubble response, while the other pulse is a fraction of the amplitude of the first to drive a linear bubble response. When the low amplitude pulse is rescaled to match the high amplitude pulse and subtracted, the linear tissue signal is removed while the nonlinear signal remains. First proposed in 1996, PAM preserves the odd harmonic component, unlike PI sequences (Brock‐Fisher, Poland, and Rafter ¹⁹⁹⁶; Eckersley, Chin, and Burns ²⁰⁰⁵).

While numerous other techniques for contrast pulse sequencing have been developed and continue to be developed, PAM and PI nonlinear contrast modes are readily available and commonly implemented on commercial US systems. A narrative review of CEUS through pulse sequencing has recently been published and is recommended (Bisht et al. 2023).

3.3. Nanobubble‐US Interactions

Like MBs, when exposed to US waves, NBs undergo oscillation due to the pressure changes of the incident wave. The interaction of US with a gas‐filled bubble is dependent on its radius, and thus NBs are predicted to resonate at higher frequencies than MBs. NBs ranging in diameters of 0.2–0.8 μm may have resonance frequencies (f_r) ~30–250 MHz depending on shell properties, which lies beyond the clinically employed frequency range (Helfield, Zou, and Matsuura ²⁰²¹). Thus, the NBs act as strong resonators of US waves at sufficiently high frequencies. However, for clinical applications, NBs are being employed under off‐resonance conditions. In such scenarios, the nonlinear response drives the signal, and the US exposure parameters (frequency and pressure) determine the degree of NB nonlinear oscillations and, therefore, contrast enhancement. Recent work has elucidated more on these complex relationships. The following sections outline studies explored for improved NB‐CEUS image quality and the mechanisms which led to their improvement.

3.3.1. Frequency‐Dependence

Considering the size range of biomedical NBs (100–500 nm), most medical imaging applications involve off‐resonance (< f_r) NB interactions. However, off‐resonance, frequency‐dependent NB response has been demonstrated. Kumar et al. investigated the effect of transducer operating frequency on acoustic contrast generated by an equal number of NBs (640 nm) and MBs (1300 nm) in tissue‐mimicking Doppler flow phantoms and found the contrast signal generated by NBs was ~16.3% lower as compared to MBs when the transducer was operated at a central frequency of 5 MHz. However, the difference in signal was narrowed down to nearly equal contrast when imaged at 6 MHz (Kumar et al. 2022). At these driving frequencies, greater backscatter from MBs would be expected since an equal number of MBs and NBs still results in 3 orders of magnitude higher gas volume in the MB sample. And yet, despite being driven far off resonance, even slight shifts in driving frequency can generate sufficient nonlinear oscillations from NBs for effective contrast enhancement. This study elucidates how selection of driving frequency can have an impact on nonlinear signals from NBs, even off‐resonance. Frequency dependence of NB backscatter has also been explored at slightly higher frequencies.

Capitalizing on frequency‐dependent responses can lead to unique imaging approaches. Recently (Karlinsky et al. 2023) used the frequency‐dependence of NB (and MB) behavior to produce a new imaging scheme based on mixed‐frequency PI imaging. Using a broadband linear array transducer, a dual‐frequency pulse (4 MHz + 8 MHz) was engineered where one full frame was produced from coherent compounding with 9 steered plane waves of single‐cycle pulses, first with positive and then with negative polarity. From their imaging scheme (4 MHz + 8 MHz), they achieved 8.4 dB contrast improvement over single‐frequency pulses (4 MHz) for NB imaging within the clinical frequency range (Karlinsky et al. 2023).

Ultimately, depending on the driving frequency employed, the response from NBs may differ in efficacy. Since NBs are typically more effective for contrast enhancement at higher frequencies, NBs broaden the relevant frequencies for CEUS imaging. Meanwhile, clinical applications of high‐frequency US imaging are growing (Lockwood et al. 1996; Mlosek, Migda, and Migda ²⁰²⁰; Russo et al. ²⁰²²). High‐frequency US (> 15 MHz) is increasing in popularity but suffers from limited penetration depth and high background signal. MBs are less effective as contrast agents as driving frequency increases further off‐resonance (> 1–6 MHz), requiring high driving pressures for sufficient MB activity. At such frequencies, NBs are increasingly relevant as driving frequencies approach the resonance frequency of NBs. Frequency is only one component contributing to the acoustic response. Next, we discuss the important effects of pressure on the bubble response.

3.3.2. Pressure‐Dependence

Pressure‐dependence of bubble signal can take on two forms: (1) a shift in resonance frequency or (2) a shift in nonlinearities of the NB oscillation. Pressure‐dependence is attributed to the asymmetry of the bubble contraction and expansion during the volumetric oscillation of the agents and impacts the volumetric behaviors differently near or far from the agent's resonance frequency. The pressure‐dependent resonance of MB agents is a well‐documented phenomenon. Increasing the pressure on the agents, whether the ambient pressure or the insonation pressure, decreases the frequency at which the agent experiences resonance. This phenomenon has been shown at both low (Gong, Cabodi, and Porter ²⁰¹⁰) and moderate (Doinikov, Haac, and Dayton ²⁰⁰⁹) driving pressures. For instance, Gong et al. demonstrated a ~28% decrease in resonance frequency for large bubbles (3.4–12.5 μm diameter) from a 50 kPa increase in insonating pressure (from 10 kPa; linear oscillation regime) (Gong, Cabodi, and Porter ²⁰¹⁰). This effect was less pronounced at higher pressures, where it was seen that increasing from 100 to 200 kPa (nonlinear oscillation regime) for a 1.3 μm bubble resulted in only ~8.3% decrease in resonance (Doinikov, Haac, and Dayton ²⁰⁰⁹). However, it is unclear whether the effect is less pronounced due to bubble dynamics owed to the material properties of the shell, as suggested by the authors (Doinikov, Haac, and Dayton ²⁰⁰⁹), or if the high polydispersity of the bubble population (Definity) played a role in the experimental findings.

NBs have also been shown to experience pressure‐dependent resonance. In a theoretical examination of the resonance frequency of NBs (~400 nm diameter), the resonance frequency of the lipid‐shelled NBs (modeled by low shell surface tension) were shown to shift from ~80 to < 50 MHz with an increase in pressure from 10 to 50 kPa (JafariSojahrood et al. ²⁰¹⁷). Moreover, with a further increase in pressure to 200 kPa, a further decrease in the resonance frequency was observed. Owing to pressure‐dependent resonance and the lower surface tension of flexible lipid shells, NBs are capable of being effective scatterers at clinically‐relevant frequencies (JafariSojahrood et al. ²⁰¹⁷; Sojahrood, Earl, et al. ²⁰²¹).

Off resonance, NBs are still highly sensitive to the driving pressure of the incident wave. Pellow et al. investigated the dynamics of porphyrin‐encapsulated NBs (number distribution mode 0.24 μm, volume distribution mode 0.28 μm) at clinically relevant frequencies (2.5 and 8 MHz) as a function of pressure (0.1–1.0 MPa; see Figure 3). They showed that NBs can suddenly initiate nonlinear scattering at clinically relevant frequencies with a pressure‐dependent onset (Pellow et al. 2018). By comparison, their MB population (number distribution mode 0.21 μm, volume distribution mode 3.9 μm), over the same parameters exhibited a gradual pressure‐dependent signal increase, demonstrating reduced pressure sensitivity. They also investigated NB nonlinear behavior at high frequencies (12.5, 25, 30 MHz) with different pulse types in single‐ and multi‐pulse sequences to examine behavior under conditions relevant to high‐frequency imaging (Pellow et al. 2021). They demonstrated that porphyrin NBs can initiate nonlinear scattering at high frequencies in a pressure‐threshold dependent manner, as previously observed at low frequencies (Pellow et al. 2021).

FIGURE 3.

Representative relative received (Rx) power spectra at 2.5 MHz (a)–(d) transmit frequency over a range in pressures. The top row depicts agent behavior in the vessel phantom, and the bottom row illustrates received spectra for agents embedded in the tissue phantom. Columns alternate between microbubble and nanobubble samples. Data are averaged over the first five ultrasound pulses. Reproduced from Pellow et al. (2018) with permissions pending.

Pressure dependencies of NBs have also been elucidated through precise control over size and shell structure, generating tunable acoustic responses (Jafari Sojahrood et al. 2021). Sojahrood et al. showed theoretically and experimentally that an increased shell stiffness of lipid‐shelled NBs (~200 nm) increases the pressure threshold for the sudden amplification in the scattered acoustic signal from the NBs (Jafari Sojahrood et al. 2021). Together, these results (see Figure 4) indicate that the nonlinear behavior of NBs at low and high frequencies can be sufficiently high for imaging contrast, provided sufficiently high driving pressures.

FIGURE 4.

Contrast enhancement of polydisperse NB solution with (a) flexible and, (b) stiff shells, and of filtered monodisperse NB solution with (c) flexible, and (d) stiff shells relative to the agarose phantom for different PNPs. Arrows mark the pressure threshold (P _t) of sudden signal enhancements exhibited by monodisperse populations. Representative ultrasound contrast harmonic imaging mode images of solutions of filtered monodisperse: Flexible and stiff shell NBs for different PNPs corresponding to (c) and (d). Adapted from Jafari Sojahrood et al. (2021) with permissions requested.

3.3.3. Polydispersity and Multi‐Bubble Interactions

Models for bubble‐US interactions typically consider the interaction of an acoustic wave with a single bubble (Versluis et al. 2020). In practice, however, the bubble behaviors of these models are only realized at sufficiently low bubble concentrations such that the inter‐bubble spacing is large enough that agents are not excited by their nearest neighbors. As bubble concentrations increase to clinically relevant quantities, the average proximity of an agent to its nearest bubble neighbor makes inter‐bubble effects non‐negligible. Attempts to model bubble‐bubble interactions are usually models involving two agents. In such models, the concentration of the population is typically reflected by the inter‐bubble distance, where closer proximity of an agent to its nearest neighbor is higher for populations of higher concentration. Meanwhile, the polydispersity of the sample population is controlled by the size difference between the two agents in the model, where the greater the size difference, the greater the bubble polydispersity. These assumptions significantly simplify the modeling needed to investigate such effects.

For context, the maximum dose for clinical applications of Definity injection is 20 μL/kg. Using the Nadler equation to estimate the blood volume of the average adult American woman (height = 5′4″; weight = 77 kg (Center for Disease Control FastStats ²⁰²¹); blood volume 4.232 L) results in a final concentration of ~3.56 * 10⁶ MBs/mL within the body, and a corresponding maximum bubble‐bubble distance of ~92 μm (center‐center; calculated from the volume per MB for Definity). In preclinical studies using NBs, higher concentrations of up to 4 orders of magnitude are commonly used, resulting in a shorter inter‐NB distance than in MB populations. At such distances, each agent may be driven by the combination of incident acoustic wave and acoustic re‐radiation from neighboring agents. Since one of the most significant benefits of NB formulations is the ability to produce agents with high yield (~10¹¹–10¹²), especially compared to MB formulations (~10⁸–10¹⁰), inter‐bubble interactions may play an important role in NB nonlinear US imaging.

Beyond inter‐agent proximity, the size differences between the neighboring agents can vary significantly. Commercially available MB populations are highly polydisperse, as shown in Figure 2. For instance, number‐weighted distributions of commercially available agents often show that only ~5% of the agents in the population are 1 μm or greater. Recent monodisperse MB populations have narrowed this range through several techniques, including pore filtering, sorting by size (Kok, Segers, and Versluis ²⁰¹⁵) or acoustic property (Segers and Versluis 2014), or microfluidic monodisperse agent production (H. Lin, Chen, and Chen ²⁰¹⁶). Segers et al. achieved a polydispersity index (ratio of standard deviation to mean radius) of 9% after acoustically sorting from a commercially available agent with a native polydispersity index of 60% (Segers et al. 2018). In practice, however, commercially available agents can range from 0.1 to 15 μm, representing a nearly 150× difference in diameter between the smallest to largest bubbles in the population. By comparison, recent polydisperse NB formulations have been reported with sizes 125–645 nm (5× difference between smallest to largest bubble diameter) and monodisperse populations 105–375 nm (3× difference between smallest to largest bubble diameter) (Jafari Sojahrood et al. 2021). These ranges need to be considered in models.

Numerous models have explored inter‐MB interactions under US stimulation (Haghi, Sojahrood, and Kolios ²⁰¹⁹; Haghi and Kolios ²⁰²²; Kagami and Kanagawa ²⁰²³; Kikuchi, Kanagawa, and Ayukai ²⁰²³). From Dzaharudin et al., it was found that as MBs were clustered closer together, their oscillation amplitude for a given applied US power was reduced, and for spacing < 10 radii, nonlinear subharmonics and ultraharmonics were eliminated (Dzaharudin et al. 2013). They also found that the proximity of a larger bubble in a cluster dominated the cluster dynamics and suppressed response from surrounding smaller agents. It was also shown recently that the larger MB in a cluster can control the dynamics of the smaller MB and force the smaller MB to exhibit the same nonlinear behavior (e.g., 1/2, 1/3, 1/4 subharmonics) as of the larger MB (Jafari Sojahrood et al. 2016).

The number of studies of bubble‐bubble interactions for NBs is limited. The most inclusive modeling range was investigated in (Yusefi and Helfield 2022) and spanned from 0.5 to 4 μm to cover partial NB and traditionally reported MB size ranges. They studied bubble center‐to‐center distances varying from 2 to 32 μm at low pressures (1–120 kPa). These results from Yusefi and Hellfield show that when a given MB approaches another MB of the same size (i.e., monodisperse population), each bubble experiences a decrease in resonant frequency (~7%–10%) and an increase in max amplitude (9%–11%). Meanwhile, large bubbles were shown to activate smaller bubbles at the resonant frequency of the larger bubbles, thus broadening the range of acoustic activity of agents off‐resonance (Yusefi and Helfield 2022).

No studies to date have shown the effects of the smaller bubble on the large bubble when driving closer to the resonance of the smaller bubble. Additionally, the smallest bubble investigated by (Yusefi and Helfield 2022) was 500 nm in diameter, which is only reported as the minimum agent size due to the detection limits of the measurement apparatus used on commercially available agents such as Definity. Commercially available agents have been under‐characterized at small size ranges. Thus, there is currently no appropriate study indicating the response from NB clusters at higher frequencies. Since high‐frequency imaging is gaining interest in the field, it would be beneficial to understand the effects NBs may have on their resonance on a larger agent, especially considering their large concentration.

Because diagnostic US scanners typically operate over a narrow frequency bandwidth with respect to that of the resonance frequencies of a typical US contrast agent, only a small fraction of the bubbles in a polydisperse agent resonate to the driving US pulse and thereby contribute to the nonlinear echo signal. Experimentally, monodisperse populations have been shown in vitro to produce more specific and higher amplitude backscatter response to US stimulation (Helbert et al. 2020; Segers et al. ²⁰¹⁸). For instance, in the 2021 study by Sojahrood et al., higher sensitivity to driving parameters is demonstrated from monodisperse populations than polydisperse populations (Figure 4) (Jafari Sojahrood et al. 2021).

4. Image and Signal Processing Techniques

Many of the image and signal processing techniques currently in use for MB US imaging translate to NB imaging. Multiparametric dynamic contrast‐enhanced US (dCEUS) imaging was developed using MB contrast agents to quantify tissue perfusion (Fröhlich et al. 2015). In these techniques, quantities extracted from time‐intensity curves (TIC), such as the wash‐in slope and the area under the curve, representing blood flow or blood volume, are used to detect and measure changes in vascularization. Several reviews discuss preclinical and early clinical dCEUS applications (Fröhlich et al. 2015; Quaia ²⁰¹¹; Selby, Williams, and Phillips ²⁰²¹). Standardizing methods of measurement, analysis, and interpretation of results from dCEUS on treatment monitoring in oncology has been an ongoing interest (Dietrich et al. 2024). However, the small size of NBs allows access to smaller capillaries and deeper tissue penetration than MBs, resulting in kinetics which occur on much slower timescales. Moreover, their size and concentration suggest that the contrast in an imaging voxel is likely from the combined contribution of the scattering of many NBs in clusters. dCEUS using NBs requires cautious interpretation and, in some cases, unique analytical techniques. The following section discusses successes and limitations of dCEUS using NBs and highlights novel techniques unique to NB use. In dCEUS, after injecting agents into the bloodstream, TICs can be obtained by measuring and averaging the US signal from a location in the image representing the agent concentration as a function of time. Interpreting the shape of these averaged curves has been effective for both MB and NB imaging and has even elucidated dynamics unique to NB biodistribution. For instance, extracted TICs from MBs usually exhibit a strong, rapid first pass and a second, relatively weaker recirculation pass of the injected MBs within a time range of 1 min (Mischi, Kalker, and Korsten ²⁰⁰⁴). NBs, like MBs, exhibit a rapid first pass. Unique to NBs, a second, stronger recirculation pass may occur within 3–15 min postinjection (C. Chen, Perera, Kolios, Wijkstra, Exner, et al. ²⁰²²). This second wave is thought to result from in vivo kinetics of NBs within the internal vasculature and organ structures.

Commonly averaged over a larger region of interest, the TIC can be further assessed for differences in energy (area under the curve), tissue access (time of arrival), washout (signal decay after peak intensity) and other parameters. In the presence of leaky vasculature, NBs have repeatedly been shown to contribute to higher area under the curve, high SNR/CNR for an extended time, prolonged time to peak, and a greater degree of tumor filling compared to MBs. In dCEUS, these parameters are typically used to establish accumulation. Quantification can be performed by recording the replenishment intensity time course of the imaging plane after the local disruption of agent during a constant infusion (Hudson, Karshafian, and Burns ²⁰⁰⁹). These parameters from NB‐CEUS can be compared to other agents to identify specific biomarkers. For instance, a comparison of NB parameters to size‐isolated MB agents can be used to establish vessel boundaries, while a comparison of targeted NBs to nontargeted agents is used to separate NB access effects from NB accumulation effects.

While much of the analysis has been performed by averaging regions of interest in image acquisitions, the tumor microenvironment has long been known to be highly heterogeneous. The capture of tumor heterogeneity is a topic of interest since the tumor microenvironment heterogeneity also contributes to treatment resistance. Pixelwise analysis is of particular interest as it preserves the spatial heterogeneity of the tumor at the resolution level of the acquisition technique. Previous groups have demonstrated the possibility of assessing the TICs in NB CEUS at the pixel level (~ 100 μm), resulting in the generation of parametric maps (C. Chen, Perera, Kolios, Wijkstra, Exner, et al. ²⁰²²). Chen et al. produced pixel‐wise analysis on animal models by fitting pixel TICs using pharmacokinetic models and subsequent parametric mapping of extracted values postfit. The models used suggest that the second‐wave phenomenon is a combination of the fast NB transport in large vessels and the slow NB perfusion of the capillaries and interstitial spaces (C. Chen, Perera, Kolios, Wijkstra, Exner, et al. ²⁰²²).

Owing to the unique capacity for extravasation by NBs, a new image processing technique was reported recently which shows NB accumulation within a tumor in a single image (Wegierak et al. 2024). A pixel‐wise parametric map was developed using measures of randomness in pixel TICs to isolate tumor tissue from healthy tissue based on the long timescales of NB accumulation within a tumor. This new technique, coined decorrelation time mapping, is the first known postprocessing method designed to extract and rapidly validate NB kinetics. Since NB motion through the vascular network is expected to be rapid relative to extravasated agents moving through the interstitium, fluctuations in TIC signal are more rapid for voxels capturing the vascular network, causing short decorrelation times associated with signal from intravascular NBs (Wegierak et al. 2024). In their study, decorrelation time maps compared agent accumulation in PC3 tumors in a mouse model. Their results also show sensitivity to targeting technique (passive vs. active), suggesting that this can be used as new method for measuring targeting efficiency. Since MBs are intravascular agents subject to higher flow rates and are less effective at molecular targeting than NBs, this technique is uniquely tailored to NB‐CEUS. Currently, this technique is the only known NB‐specific approach for mapping and interpreting CEUS acquisitions.

More advanced NB US imaging analysis methods have also been developed. Pharmacokinetic modeling involves compartment modeling relating the temporally dependent concentration of an agent to the temporal US signal intensity (Turco et al. 2020). By fitting the model to measured TICs, parameters are extracted to produce quantitative measures of the tissue and its properties. This has been applied to both contrast and Doppler (Seidel, Beller, and Kaps ¹⁹⁹⁶) US imaging modes. These methods have also been applied to NB US imaging to quantify extravasation in prostate cancer (C. Chen, Perera, Kolios, Wijkstra, Mischi, et al. ²⁰²²; H. Wu et al. ²⁰¹³).

5. Preclinical Studies and Future Clinical Applications

Thus far, no NB formulation has been approved for clinical use, representing a notable gap in their translation to practical medical applications. Nevertheless, extensive investigations have been undertaken to evaluate the clinical potential of NBs, with a predominant focus on their utility in oncology. Nononcological applications of this technology are also being actively explored. The applications unique to NBs are enabled by their ability to extravasate. This applies broadly to inflammation applications, including cancers, diabetes, and more tissue pathologies. However, investigations of NBs outside oncological applications have been limited since stable NB formulations have only been developed in the last decade. In studies comparing MBs to NBs, NBs have demonstrated improved access and retention in tissue regions inaccessible by MBs. This has been shown in numerous disease types (cancer, diabetes, stroke), cancer types (prostate, breast), and other models (microfluidic, mouse, rabbit & dog), with and without targeting moieties. Across models NBs have provided improved targeting efficiency to diseased areas because of their small size. These studies are briefly discussed here.

5.1. Oncological Applications

Unlike MBs, which play a crucial role in vascular imaging (Lyshchik et al. 2018), NBs, due to their size, leverage the permeability of tumor vasculature (due to chronic inflammation), enabling targeted delivery to cancer cells beyond the blood vessels and accessing the tumoral microenvironment (H. Wu et al. ²⁰¹⁹). Owing to their enhanced access beyond the vascular network, they also offer a more accurate assessment of tumor heterogeneity, offering unique opportunities in US applications for oncology. Thus, the oncology applications presented here have comprised most in vivo NB CEUS imaging studies.

Research in mouse tumor models has focused on numerous cancer types, including prostate and breast cancers, and ovarian, thyroid, and gastric carcinomas (Fan et al. 2013; Y. Gao et al. ²⁰¹⁷; Q. Jiang et al. ²⁰¹⁶; R. Perera et al. ²⁰¹⁹; Y. Wang, De Leon, et al. ²⁰²¹; Xie et al. ²⁰²²; Y. Zhu et al. ²⁰²⁰). Decorating NBs with targeting epitopes that bind to overexpressed biomarkers on tumors has enabled molecular US imaging of cancer in animal models. For instance, Gao et al. demonstrated the application of CA‐125‐targeted NBs imaged with contrast harmonic US for imaging of CA‐125 positive OVCAR‐3 tumors in mice (see Figure 5). Surface functionalization of the NBs with the antibody achieved rapid tumor accumulation, higher peak intensity and slower wash out rates in OVCAR‐3 tumors compared to CA‐125 negative SKOV‐3 tumors (Y. Gao et al. ²⁰¹⁷). As such, tailored approaches to the cancer type typically involve adjusting the targeting epitope integrated onto the NB surface to maximize tumoral accumulation.

FIGURE 5.

Demonstration of nanobubble (NB) use in contrast‐enhanced ultrasound (CEUS) imaging. (a) Adapted and reproduced with permissions requested (W. B. Cai et al. ²⁰¹⁵). (a.i) Ultrasound‐enhanced images of subcutaneous tumors before and after caudal vein injection of (top) NBs and (bottom) SonoVue at 10 s, 30 s, 2 min, and 5 min. (a.ii) Time‐intensity histogram of the tumor gray‐scale enhancement after caudal vein injection with NBs (red) and SonoVue (blue). **p < 0.01 comparison of NBs and SonoVue at the same time point. (b) Adapted and reproduced with permissions requested (Peyman et al. 2016). (i) High Frequency Ultrasound (US) images of mouse aorta after bolus delivery of (left to right) mixed population bubbles, NBs, or MBs via tail vein catheter at 0.6 mL min⁻¹. The aorta was identified in each mouse (circled). (ii) Size distribution of NBs as determined by particle tracking. Inset: TEM image of NBs. (c) Adapted and reproduced with permissions requested (Wei et al. 2022). (i) US imaging of gas vesicles (GVs) and lipid MBs in MB49‐tumor: a B‐mode and nonlinear contrast images of GVs and MBs in 15 min after tail injection; (ii) time–intensity curves of GVs and MBs perfused into the tumor tissue.

Prostate‐specific membrane antigen (PSMA) ligand‐marked NBs have been used in prostate cancer imaging. Perera et al. demonstrated that PSMA‐targeted NB (PSMA‐NB) could achieve twice the peak CEUS intensity of clinically approved MBs using an in‐house mouse model. Their study revealed enhanced molecular targeting, exemplified by the prolonged half‐life of PSMA‐NB compared to untargeted counterparts in a PC3 xenograft model (R. Perera et al. ²⁰¹⁹). Moreover, when the tumor models were injected with MBs only the tumor periphery were perfused, indicating limited MB access. Similar success of NBs was reported by Wang et al. in an orthotopic prostate cancer model (Y. Wang, De Leon, et al. ²⁰²¹). Additionally, Zhu et al. introduced a dual imaging approach with PSMA‐NB and superparamagnetic iron oxide nanoparticles (SPIONs) in an LNCaP mouse model, showcasing promising results in MRI (Y. Zhu et al. ²⁰²⁰).

In breast cancer imaging, Li et al. utilized a fluorescent NB ligand for neuropeptide Y, selectively enhancing four T1 cells in subcutaneous mouse xenografts (J. Li et al. ²⁰¹⁷). They established high affinity to tumor cells with maximum accumulation occurring in the tumor 3 min postinjection. In another approach, Du et al. developed a dual‐targeted nanoscale contrast agent for breast imaging, demonstrating tumor enhancement, and the superiority of dual‐targeted bubbles over untargeted ones (Du et al. 2018). Their study identified the benefits of dual targeting compared to single‐ or passive‐targeting using US.

Beyond prostate and breast cancers, NB applications extend to various solid tumors. Zhu et al. introduced a carbonic anhydrase IX (CAIX)‐targeted NB with heightened intensity and prolonged signal duration, specifically in CAIX‐positive tumors (L. Zhu et al. ²⁰¹⁷). Gao et al. focused on ovarian cancer imaging, developing a CA‐125 targeted nanoscale US contrast agent with comparable success (Y. Gao et al. ²⁰¹⁷). Numerous studies explore NB applications in diverse tumors such as thyroid carcinoma, pancreatic adenocarcinoma, and gastric carcinoma, presenting a comprehensive landscape of their potential in oncological imaging (Fan et al. 2013; Xie et al. ²⁰²²; Yang et al. ²⁰²¹).

These studies suggest that molecular targeting of NBs achieves enhanced and prolonged contrast at the tumor site facilitated by tumoral NB accumulation. Thus the ligands that have been tested for molecular targeting of NBs have been provided in Table 2. Many of the molecular targets explored above are expressed beyond the vascular network, indicating the inability to achieve comparable results with MBs. The continued identification of overexpressed biomarkers on cancerous cells will help broaden the cancer types that can benefit from NB CEUS. However, the expression of targets on the cancer cells must be sufficiently high to reach localized critical agent concentrations for contrast enhancement.

TABLE 2.

Summary table of nanobubble formulations produced for ultrasound imaging applications.

Production method	Shell	Core	Size	Yield [NBs/mL]	Ultrasound frequencies	Targeting/loading	References
Extrusion	Phospholipid	C3F8	160 ± 50 nm	6.2 ± 1.8 × 1010	12 MHz	—	(Counil et al. 2022)
Freeze‐dry and gas exchange of insulin‐loaded CS‐PAA nanoparticles	Chitosan polyacrylic acid	C5F12	142.2 ± 0.64 nm	N.A.	7 MHz	Insulin	(Yi et al. 2023)
Mechanical agitation and differential centrifugation	Phospholipid	C3F8	274 ± 8 nm	4.07 × 1011 ± 3.15 × 1010	7, 8, 10, 12, 18, and 25 MHz	PSMA‐1	(de Leon et al. 2019; R. Perera et al. 2019)
Mechanical agitation, differential centrifugation and filtration	Porphyrin‐lipid	C3F8	(Number‐mode) 240 nm	Native conc. N.A. > 106	1.5 and 8 MHz	—	(Pellow et al. 2018)
Mechanical oscillation and differential centrifugation	Phospholipid	C3F8	(FOL)2‐NB: 287 ± 23 nm FOL‐NB: 244 ± 35 nm Non‐Fol‐NB: 129 ± 12 nm	Native conc. N.A. > 106	9 MHz	Folate	(S. Duan et al. 2017)
Microfluidics	Phospholipid	C4F10	260 nm	> 1011	15 and 40 MHz	—	(Peyman et al. 2016)
Microfluidic reconstruction of MBs to NBs	Phospholipid	C3F8	640 nm	3.7 × 1010	5, 6, and 18 MHz	Nanobody FN3hPD‐L1	(Kumar et al. 2022)
Oil‐in‐water emulsion solvent evaporation	PLGA	C3F8	230.2 ± 58.5 nm	N.A.	22 MHz	FITC‐anti‐HER2 and PE‐anti‐VEGFR2	(Du et al. 2018)
Oscillation and centrifugation	Phospholipid	C3F8	503.7 ± 78.47 nm	10.70 ± 0.82 × 108/mL	5–12 MHz	CAIX polypeptides	(L. Zhu et al. 2017)
Temperature‐regulated self‐assembly	Phospholipid	SF6	1.68 ± 0.11 μm, 704 ± 7 nm, and 208 ± 6 nm	2.06 ± 0.9 × 109	18 MHz	—	(J. Jin et al. 2020)
Thin‐film hydration and self‐assembly following mechanical vibration	Phospholipid	SF6	206.6 ± 1.6 nm	1.46 × 1010	18 MHz	S1P and HMGB1 direct inhibitor DG	(X. Wang, Liu, et al. 2023)
Thin‐film hydration	Phospholipid	C3F8	565.2 ± 201.5 nm 457.9 ± 113.8 nm	107	5 MHz	—	(W. B. Cai et al. 2015)
Thin‐film hydration	Phospholipid	C3F8	478.2 ± 29.7 nm; 349.8 ± 159.1 nm	Native conc.s N.A.	7.5 MHz	Anti‐ErbB2; IR‐780 iodide and docetaxel	(Yang et al. 2015); (Yang et al. 2021)
Thin‐film hydration	Phospholipid	PFC	151 ± 3 nm	Native conc.s N.A.	14, 30, and 40 MHz	Calcein, Nile Red	(Hanieh et al. 2022)
	Lipid‐DCP		130 ± 2 nm
	Span‐20		173 ± 4 nm

Note: Some values not available (N.A.) from the original publications.

5.2. Non‐Oncological Applications

The scope of NB‐CEUS extends beyond oncological applications. NBs have been investigated for their utility in inflammatory conditions characterized by enhanced vascular permeability. Ramirez et al. employed NBs to evaluate inflammation in pancreatic islets, serving as a potential prognostic indicator for type‐1 diabetes in nonobese diabetes (NOD) mouse models (Ramirez et al. 2020). Their study demonstrated a heightened CEUS signal in the pancreas of NOD mice compared to controls, suggesting the potential for early diagnosis and monitoring of inflammatory processes in pancreatic tissues (Ramirez et al. 2020). In the same study, MBs could not evaluate inflammation since they exhibited no prolonged signal beyond initial vascular filling after infusion.

Additionally, investigations have extended to cardiac and liver grafts posttransplantation. Liu et al. established a positive correlation between T‐lymphocyte‐targeted NBs and T‐lymphocyte accumulation in heart allografts, providing a predictive marker for acute organ rejection. In the context of liver transplantation (Xie et al. 2016), Xie et al. explored the use of intracellular adhesion molecule‐1 (ICAM‐1) targeted NBs to predict ischemia/reperfusion (I/R) lesions. Their study demonstrated elevated retention of targeted NBs in rabbits with I/R injury, underscoring the potential of NBs in assessing posttransplantation tissue damage (Xie et al. 2016).

Beyond these applications, ongoing research is actively evaluating NB efficacy in diverse conditions such as diabetic cardiomyopathy, Alzheimer's disease, and stroke (J. Gao et al. ²⁰²¹; Y. Lu et al. ²⁰¹⁶; Mi et al. ²⁰²²). Lin et al. prepared chitosan polyacrylic acid NBs for sentinel lymph node identification via CEUS (Y. Lin et al. ²⁰²²). Owing to the small size of the agents, NBs are capable of unique interactions with cells, including internalization (R. H. Perera et al. ²⁰²²), hitch‐hiking (M. B. Cooley, Abenojar, et al. ²⁰²³), and targeting (Ding et al. 2016; Mi et al. ²⁰²²; R. Perera et al. ²⁰¹⁹). Recently, Jiang et al. produced natural killer (NK) cells and labeled them with GVs for adoptive immune cell tracking (Y. Jiang et al. ²⁰²³). In doing so, they were able to capture the dynamics of NK cell trafficking to the tumor site without impediment to cell access. This may be a potential future avenue to explore with NBs. Investigations directed by NB‐body interactions will expand the relevant disease targets.

5.3. Nanobubble Trafficking and Biodistribution

NBs, like their larger counterparts, are currently used in various in vivo applications. Gaining a comprehensive understanding of their biodistribution is essential, especially as NBs are multifaceted agents with both imaging and therapy capabilities. However, for both MBs and NBs, the number of pharmacokinetic studies is limited.

Upon initial injection, MBs and NBs are intravascular agents. After injection into the bloodstream, MBs circulate for only a few minutes (~2–5 min) before being cleared through dissolution and phagocytosis (Stride et al. 2020). The pharmacokinetics of heavy gases (e.g., C₃F₈, C₄F₁₀, and SF₆) delivered to the body using MBs have been determined by measuring expired air and blood samples for gas presence using GC/MS methods (Landmark et al. 2008; Morel et al. ²⁰⁰⁰). It is now generally accepted that the gas core of bubble contrast agents is expired during exhalation. MBs also break down into several nanoprogeny, including NBs (Paknahad et al. 2023; Pellow et al. ²⁰²⁰), nanoparticles (Hou et al. 2020; Huynh et al. ²⁰¹⁵) and nano‐sized shell fragments/vesicles (Ho et al. 2021; Rajora et al. ²⁰²⁴). Each of these nanoprogeny can quickly diffuse through the vasculature into the interstitial space, facilitating a broader long‐term biodistribution of the agents' constituents into the body. Once any nanoprogeny has eliminated its gaseous core, its clearance profiles are more consistent with those of lipid nanoparticles (Rajora et al. 2024). Numerous other groups have studied lipid nanoparticle accumulation and pharmacokinetics and will not be reviewed here (Di et al. 2022; S.‐D. Li and Huang ²⁰⁰⁸; Luo et al. ²⁰¹⁹; Mukai et al. ²⁰²²; G. Wang, Zannikou, et al. ²⁰²¹).

Recently, Rajora et al. conducted the first longitudinal (0–48 h) multimodal pharmacokinetic study of lipid MB and nanoprogeny shell fate (Rajora et al. 2024). With NBs being one of the potential nanoprogeny of MBs, the study by Rajora et al. is relevant to understanding NB biodistribution and pharmacokinetics. In this work, Definity‐like formulations (matched size distribution, similar shell composition, and similar yield) were produced. Anionic and neutral compositions were compared to short (C16) and long (C18) phospholipid chain length compositions. It was shown that lipid‐shelled MBs are fully eliminated from blood circulation 24–48 h postinjection through hepatobiliary/fecal clearance. Within the postinjection timeline, shell clearance timeframes were shown to be three orders of magnitude longer than MB core dissolution rates (~3 min vs. 48 h). During shell clearance, Rajora et al. found that nanoprogeny accumulation occurs predominantly in the liver and the spleen, which is consistent with the fate of most nanoparticles; shells composed of short‐chain phospholipids yield preferential uptake by the liver, while formulations with long‐chain lipids exhibit preferential uptake by the spleen.

There are several critical caveats when translating MB pharmacokinetics and biodistribution studies to NBs. While the study by Rajora et al. did not identify the proportions of the various nanoprogeny after disrupting the MB agents, bulk NB populations are produced with orders of magnitude higher concentration (10¹¹) over most MB populations or any NB subpopulations in MB formulations. Moreover, when comparing injection doses with matched gas volumes—standard practice for comparison of image quality between MBs and NBs in CEUS—the total shell material amount is higher for NBs than MBs. Since clearance timeframes likely have concentration dependencies, with higher lipid concentrations requiring longer time for clearance, gas‐matched NB clearance time may be longer than MB clearance time. However, the role of agent concentration was not explored here and may be an area of future work.

Since NBs are one of the possible nanoprogeny of lipid‐MBs and the nanoprogeny of NBs are likely to overlap with those of MBs, their clearance is likely similar, but extended for NB formulations. NB formulations have been produced with circulation times extended over MB formulations by up to an order of magnitude (half‐life as long as 32 min; Batchelor et al. ²⁰²²). While most of the perfluorocarbon from MBs is exhaled within minutes (Schutt et al. 2003), the well‐established longer dissolution times of NBs under US means the prolonged presence of perfluorocarbons in the body. For instance, the biodistribution of gaseous NBs was previously assessed via CEUS, demonstrating US signal enhancement in the kidney, liver, spleen, heart, and kidneys by NBs within moments of injection and lasting up to 10 min (H. Wu et al. ²⁰¹⁹; Zheng et al. ²⁰¹²). Thus, the high stability of NBs enables their accumulation into off‐target organs before dissolution is completed. Further investigation is needed to understand the clearance of the gaseous agents at off‐target sites.

Finally, significant efforts have been dedicated to improving disease specificity, retention, and signal longevity of NBs by surface functionalization with targeting ligands (C. Chen, Perera, Kolios, Wijkstra, Mischi, et al. ²⁰²²; S. Duan et al. ²⁰¹⁷; Y. Gao et al. ²⁰¹⁷; Q. Jiang et al. ²⁰¹⁶). While critical to precision medicine, these modifications will alter the distribution and clearance pathways. For example, targeting moieties on the surface of NBs (R. H. Perera et al. ²⁰²²) makes endocytotic pathways more likely for NBs than for MBs. Size phagocytosis is predominantly for the filtration of particles > 0.5 um; polydisperse MB populations are more likely to experience phagocytosis than NBs. Additionally, while many ligands exhibit enhanced tumor targeting, many studies fail to illustrate the trafficking of targeted NBs to off‐target sites. For instance, PSMA NBs exhibit increased accumulation and retention in tumors, yet they also demonstrate heightened kidney retention over untargeted agents (R. Perera et al. ²⁰¹⁹). More effort is needed to examine the role of targeting epitopes on the final NB organ uptake and total bioaccumulation in the body.

5.4. Nanobubbles in Therapeutics: From Drug Delivery to Immunotherapy

The known advantages of NBs are consistent with the advantages of all nano‐theranostic agents: improved targeting efficiency in molecular techniques, greater surface area to volume ratio for agent loading efficiency in theranostics combined with the ability to extravasate enabling deep tissue access. The enhanced permeability and retention (EPR) effect is highly variable, but when present, significantly improves nanotherapeutic delivery in patients (H. Lee et al. ²⁰¹⁷; Ramanathan et al. ²⁰¹⁷; van der Meel et al. ²⁰¹⁹). Until the advent of NBs, these capabilities, only realized by non‐buoyant nanoparticles, were inaccessible with US technologies. These points are discussed in brief here, however, work applying NBs to US therapy is extensive and merits its own review. The reader is directed to (Dehariya et al. 2023) for more information.

Beyond imaging, US can be applied to echogenic bubbles as a noninvasive on‐demand drug delivery system (Cavalli, Soster, and Argenziano ²⁰¹⁶; Chandan, Mehta, and Banerjee ²⁰²⁰). NBs demonstrate remarkable drug carrier and theranostic capabilities attributed to their submicron‐size, high compressibility, low density, and distinctive interactions with US (Exner and Kolios 2021; J. Jin, Yang, et al. ²⁰²²). They have been used in the delivery of several therapeutic materials, including but not limited to chemotherapeutics (Chan et al. 2020; Fang et al. ²⁰²¹; Moghaddam et al. ²⁰²⁴; Nittayacharn et al. ^{2019, 2023}; Prabhakar and Banerjee ²⁰¹⁹) and genetic materials (X. Cai et al. ²⁰²⁰; Endo‐Takahashi and Negishi ²⁰²⁰; Kida et al. ²⁰²³; Su et al. ²⁰²²; M. Wu et al. ²⁰¹⁸). These therapeutic materials can be loaded onto or within the shell, allowing for functionalization and targeting with high efficiency (Batchelor et al. 2021). NBs can deliver a greater dose of the chemotherapeutic agent compared to MBs directly to the tumor tissue while reducing systemic toxicity due to the increased loading capacity, and deeper penetration (Nittayacharn et al. 2023). NBs have also been shown to have an improved safety profile over MBs owing to the improved tissue access causing reduced off‐target effects (Nittayacharn et al. 2024). Upon reaching the targeted site, acoustic waves are used to facilitate drug release through thermal/mechanical effects driven by cavitation phenomena or radiation forces (Schroeder et al. 2009).

Beyond drug delivery applications, NBs can serve as a drug‐free theranostic agent. During insonation, NBs experience cavitation that involves several mechanical and physical changes, including volumetric expansion, compression, oscillation, fragmentation, coalescence, dissolution, and/or abrupt collapse (S. Lu et al. ²⁰²²). These alterations induce a temporary and reversible disruption of nearby surfaces, enhancing permeability across barriers such as the blood–brain barrier (Gattegno et al. 2023; Yan et al. ²⁰²¹), vascular wall, and cell membrane (Helfield et al. 2016). Referred to as sonoporation, this phenomenon facilitates the extravasation of released or co‐administered therapeutic materials into the targeted region. Users may desire the use of NBs over MBs in such applications to enable strategic positioning of agents before eliciting bioeffects beyond the vascular network. The challenges in US‐induced therapy include optimizing various carrier types and understanding how they interact with biologically active substances and US. Tailored to the specific application, US triggering mechanisms utilize either high frequencies (> 500 kHz) or low frequencies (20–500 kHz), interacting with the tissue to generate thermal effects, mechanical effects such as cavitation, or a combination of both (de Leon et al. ²⁰¹⁸).

NBs have been explored as vascular‐disrupting agents, generating similar effects as MBs on tumor vasculature causing damage to blood vessels through inertial cavitation near endothelial cells (Hysi et al. 2020). Studies have shown that using NBs in conjunction with unfocused low‐frequency US can sensitize tumors and improve radiation therapy (Hysi et al. 2020; R. H. Perera et al. ²⁰¹⁴). The bioeffects of bubble cavitation within particular extravascular target cells can be concentrated using molecular targeted NBs. When combined with low‐frequency US, the combination serves as a safer alternative to extracellular and intravascular cavitation (R. H. Perera et al. ²⁰²⁴). Moreover, this approach eliminates the need for US or magnetic resonance imaging guidance, facilitating clinical translation by improving safety, accessibility, and simplifying the methodology.

Another potential therapeutic approach of NBs is sono‐dynamic/immunotherapy (J. Li et al. ²⁰²²). Sonodynamic therapy is a non‐thermally related therapeutic US application encompassing the induction of apoptosis. It can be combined with chemotherapy, and gene therapy (Tachibana, Feril, and Ikeda‐Dantsuji ²⁰⁰⁸). The acoustic cavitation of NBs incorporating photosensitizing agents like porphyrin or Iridium complex provides the advantage of activating tumor‐localized drug agents at greater depths compared to its close relative, photodynamic therapy (Bosca et al. 2018; Nittayacharn et al. ²⁰²²). In addition, studies have shown that NBs incorporating the sonosensitizer chlorin e6 (Ce6) can inhibit cancer cells by generating reactive oxygen species that are produced by the sonosensitizer, working in concert with PD1 blockage (Liu et al. 2022; Um et al. ²⁰²⁰; Zhao et al. ²⁰²¹). Recently, NBs have emerged as a novel immunological adjuvant in immunotherapy. Low‐intensity US‐induced NB cavitation promotes immune responses, enhancing CD8+ T cell infiltration and antitumor activity (Hu et al. 2022). Their combination with anti‐PD1 antibody‐labeled NBs achieves lasting inhibition of tumor growth, metastasis, and recurrence, offering a new option for combination therapy in patients with immunotherapy‐resistant tumors.

Overall, NBs provide a promising opportunity to improve treatment efficiency by integrating the benefits of nanomedicine and US into a single straightforward drug delivery platform. Several challenges still exist in developing NBs for therapeutic and theranostic applications despite the numerous therapeutic aspects that have been discussed. It is crucial to emphasize that advancements in therapeutic outcomes are closely linked to progress in diagnostic imaging. For instance, the recent research on imaging with NBs offers an opportunity to assess the distribution of the nanomedicine before triggering the desired bioeffects. As detection capabilities improve, real‐time therapy tracking is made feasible, enabling fast decision‐making for clinicians and opportunities for treatment plan intervention.

6. Challenges and Controversies

The are several ongoing controversies pervasive in NB‐CEUS research. The following section aims to highlight published works which address some of the concerns. These are summarized in Table 3.

TABLE 3.

Compiled collection of evidence for major claims in nanobubble research.

Evidence of…	Location in original article	DOI
Cellular internalization—confocal microscopy	Figure 4	https://doi.org/10.7150/ntno.64735
Extravasation and accumulation—intravital microscopy and ultrasound	Figure 7	https://doi.org/10.7150/thno.51316
		https://doi.org/10.1021/acs.nanolett.0c01310
Extravasation and accumulation—histology	Figure 7	https://doi.org/10.1016/j.ultrasmedbio.2019.05.025
	Figure 5	https://doi.org/10.1038/s41598‐022‐17756‐1
Lipid concentration prerequisites to avoid coalescence	Section 2.1	https://doi.org/10.1021/acs.langmuir.6b00616
Microbubble decay and fragmentation into nanobubbles	Figure 1	https://doi.org/10.1039/D3SM00380A
Shelf‐stability	Figure 5	https://doi.org/10.1016/j.nano.2022.102611
	Figure 3	https://doi.org/10.1039/C8NR08763F
Shell effects on resonance frequency shift	Figure 5	https://doi.org/10.1109/ULTSYM.2017.8092386
Structure—cryo‐electron microscopy and cryo‐tomography	Figure 3	https://doi.org/10.1016/j.nano.2022.102611
Structure—transmission electron microscopy	Figure 5	https://doi.org/10.1117/1.JBO.27.1.016501

6.1. Stability

The existence of bulk NBs has long been contended owing to the theoretically increasing Laplace pressure as function of decreasing radius. It is now known that the stability of NBs is due to a complex interplay of factors such as the protective shell formation, surface charge, diffusive shielding, and specific materials at the interface (Alheshibri and Craig 2018; Bu and Alheshibri ²⁰²¹; Nirmalkar, Pacek, and Barigou ^{2018a, 2018b}). The effects of dissolved impurities, pH, ions, surface charge, excess lipids and surfactant addition on NB colloidal stability have all been shown to play an important role in the formation and persistence of bulk NBs (Karimi, Parsafar, and Samouei ²⁰²⁴; Nirmalkar, Pacek, and Barigou ^2018b; Tan, An, and Ohl ^{2020, 2021}; Zhou et al. ²⁰²¹). As such, some industrial applications now routinely use ultrafine bubbles (see Moleaer Inc., https://www.moleaer.com/). Two key differences distinguish the NBs used for biomedical applications from other applications: (1) the use of a heavy hydrophobic gas (such as perfluorocarbons) rather than air, nitrogen, or CO₂ for typical “bulk” NBs and (2) an intentionally designed stabilizing shell comprised of lipids and/or polymers (typical bulk NBs studied in the physics literature are not intentionally coated). Beyond the theory of their formulation, carefully controlled experiments at very high frequencies unambiguously demonstrate the stability and scattering of NBs (Moore et al. 2020).

6.2. Foundations of Low Backscatter Estimates

We have discussed scattering potential of NBs in our previous publication (Exner and Kolios ^2021b). There are numerous studies which indicate to the nonlinear scattering potential of NBs—key to effective contrast enhancement in nonlinear imaging modes. Rayleigh scattering—the foundation for low estimates of NB scatter—does not capture nonlinear complexities of a compressible sphere encapsulated by a viscoelastic shell. If one assumes Rayleigh scattering, then scattering from a NB (e.g., 0.2 μm) compared to a MB (e.g., 2 μm) at clinical frequencies (1–6 MHz) is 1 million times less. Rayleigh scattering equations can accurately model solid spheres (or red blood cells). However, highly nonlinear properties emerge when modeling compressible materials (bubbles) with surfactant coatings (lipid shells), making Rayleigh scattering assumptions invalid. For example, the simplified equation for Rayleigh scattering does not capture MB “compression‐only” behaviors or other more complex nonlinear scattering behaviors (de Jong et al. ²⁰⁰⁷; Versluis et al. ²⁰²⁰). Assuming that the bubble oscillations are radially symmetric, variants of the Rayleigh–Plesset equation (e.g., Marmottant model for lipid‐coated MBs) can be used to predict the nonlinear behavior (generation of harmonics and sub‐harmonics) at low pressures (Sojahrood, Haghi, Porter, et al. ²⁰²¹). Modified Rayleigh–Plesset models show that (1) NBs scatter more than predicted from the Rayleigh approximation for MB and NB scattering (Sojahrood et al. 2020) and (2) the nonlinearities in NB oscillation are intensified even if the excitation frequency is off‐resonance (clinical frequencies), making NBs effective contrast agents. There is experimental (Jafari Sojahrood et al. 2021) and theoretical (Sojahrood 2021) evidence of this. More effects emerge when the bubble oscillations are not radially symmetric (Falou et al. 2012; Überall et al. ¹⁹⁷⁹), but these go beyond this discussion. When considering the shell properties, the lipid coating causes enhanced nonlinearity, generating harmonics and subharmonics that are easily detectable even at low frequencies (3–50 MHz) (JafariSojahrood et al. ²⁰¹⁷).

6.3. Coalescence

Since prior estimates of NB backscatter have reported theoretically low signal amplitudes, a common conclusion about NB studies is that the time‐dependent coalescence of NBs into MBs is the source of successful NB‐CEUS.

Sizing bubbles before and after imaging shows no shift in population size. Hernandez et al.'s work in 2018 provides experimental evidence that a stable formulation of NBs can be produced and used for contrast enhancement without coalescing (Hernandez et al. 2019). Their study used resonant mass measurement to monitor the bubble concentration and size change over time. Rather than an increase in diameter, which would indicate coalescence, a statistically significant decrease in the size was observed 1 h after the NBs were prepared. We have provided the key figure from their publication below (see Table 3).

Prerequisites to minimize coalescence are met in biomedical NB formulations. NB formulations for US imaging are typically produced with high lipid concentrations for the shell (as high as 14 mg/mL), necessary for minimizing bubble coalescence (Tim Segers ref. below). While studies specifically discussing NB coalescence have not been published, the high lipid concentration and the time‐dependent decrease in bubble size with time suggest that coalescence is not occurring.

In addition, NB populations at matched gas volume to MB populations do not generate the same US signal. If suspensions of NBs being imaged were coalescing into MBs, a time‐dependent increase in attenuation would occur, correlated to the coalescence rate as the gas volume joins into particles of increased scattering strength (assuming low transmit frequency). There are no reports of this effect in the literature. At equivalent shell compositions and volumes, as measured quantitatively using headspace gas chromatography/mass spectrometry, the acoustic behavior of NBs and MBs differs greatly (Abenojar et al. 2019, 2020) indicating stable differences in size between the isolated populations.

6.4. Nanobubble Visualization

Sub‐optical scatterers were shown to generate US signal beyond the vascular network while micron‐sized agents did not. Experimental evidence of NB acoustic signatures and their ability to extravasate and accumulate has been directly measured through simultaneous intravital microscopy and US (Pellow et al. 2020). This work demonstrates the ability for optically sub‐resolution US scatterers to accumulate at sufficient concentration to generate a detectable fluorescent signature while simultaneously producing US backscatter—which is only possible with nanoscale agents that have a gaseous core. Their study showed no support/evidence for the coalescence of the NBs into MBs since a micron‐scale bubble would be optically resolvable through intravital microscopy. Beyond intravital microscopy, confocal imaging of NBs in cells (R. H. Perera et al. ²⁰²²), molecular US imaging using NBs (Johansen et al. 2021; Y. Wang, De Leon, et al. ²⁰²¹), and histology staining (C. Chen, Perera, Kolios, Wijkstra, Mischi, et al. ²⁰²²; H. Wu et al. ²⁰¹⁹) show the presence of NBs in tissue.

7. Unmet Needs and Future Directions

There are many opportunities for future studies exploring the use of NBs in US imaging and image‐guided therapy. From optimization of formulation and imaging parameters to novel disease applications, there are a broad spectrum of avenues for exploration.

On a fundamental level, before the full the diagnostic and therapeutic potential of NBs can be realized, there is a significant knowledge gap in optimizing formulation parameters. Optimization is required to identify appropriate lipid concentration, shell surface charge, shell lipid and polymer components, and diameter. These formulation parameters play an important role in the NB surface interactions with their surrounding environment and have only been explored superficially. For example, some studies have shown that NBs, and other similar particles like GVs, interact with blood cells by loosely integrating into the cell curvature, possibly increasing their in vivo lifespan (M. Cooley, Pieper, et al. ²⁰²³; M. B. Cooley, Abenojar, et al. ²⁰²³). Other cellular interactions (e.g., NBs with serum proteins, or NB internalization mechanisms) are unexplored and could be exploited for applications in targeted imaging, cellular tracking, and increasing in vivo imaging time. The reader is directed to (Villanueva‐Flores et al. ²⁰²⁰) for an overview of cellular interactions with nanomaterials. These formulation parameters in conjunction with their effects on cellular interactions must be studied not only in vitro but also in vivo to maximize criteria like circulation time and echogenicity, which are critical to the success of many applications (e.g., tumor imaging and treatment).

With changes to agent, composition comes changes to harmonic response and the renewed need for optimization studies exploring appropriate US driving parameters. Pulse parameters including pressure, frequency and cycle number play a critical role in high contrast long lasting signal. Table 2 shows applications exploring NBs with transmit frequencies ranging from 1.5 to 40 MHz, however, the systematic study of exposure conditions to identify NB signal optimum has not been conducted. A thorough study of such parameters would also help with identifying the best pulse sequences for NB imaging and would have downstream benefits in both imaging and therapeutic applications.

In post‐processing, one untapped area of interest is super‐resolution US imaging, where individual MBs are traditionally tracked to produce highly detailed vasculature maps. If this technique could be applied using NBs, there is potential for smaller and pathological vessels in tumors to be mapped, improving our understanding of tumor vasculature and improving predictions of treatment efficacy.

Clinically, NBs are ideal for oncology applications due to the heightened vascular permeability of tumor vasculature. Beyond oncology, studies have been limited. Other diseases that could benefit from NB‐based imaging and therapy include ischemic stroke, cardiovascular diseases, chronic bowel disease, acute pancreatitis, kidney disease (acute, chronic, and diabetic), diabetic retinopathy, and diabetes (Britzen‐Laurent, Weidinger, and Stürzl ²⁰²³; A. Cai, Chatziantoniou, and Calmont ²⁰²¹; Weis ²⁰⁰⁸; X. N. Wu ²⁰⁰⁰). For example, NBs could be used to evaluate the initial stages and progression of disease and subsequent treatment in patients with acute pancreatitis. Beyond diagnostic evaluation of pathological vasculature, they exhibit potential for treatment by delivering drug, immunotherapy (e.g., mRNA), or even transcription factors due to their uniquely flexible shell and ability to cavitate which improves intracellular uptake (Kim et al. 2021; Kopechek et al. ²⁰¹⁵; K. Lee et al. ²⁰¹⁵; Liao et al. ²⁰¹⁵). For instance, in one unique application of NB use, Wang et al. showed that NBs can enhance visualization of the epileptic brain and deliver drugs with anti‐inflammatory and immunomodulatory properties to ameliorate seizure symptoms (X. Wang, Liu, et al. ²⁰²³). The potential for new clinical applications of NBs is broad due to the high prevalence of inflammation across diseases.

An emerging application of NBs, especially for the abovementioned applications involving pathologically permeable vasculature, is companion diagnostics. Companion nanoparticles are used to evaluate if a patient could be a candidate for nanoparticle therapy. This can be done by studying vascular parameters (e.g., vascular permeability, vessel density, perfusion, etc.), which are biomarkers that NBs are uniquely suitable for due to their ability to reach microvasculature and extravasate from hyperpermeable vessels. Other biomarkers that may offer additional insights for patient stratification include time‐intensity curve parameters, decorrelation time, and matrix properties like collagen content. NBs have already been successfully used for companion diagnostics in oncology (M. B. Cooley, Wegierak, et al. ²⁰²³; M. B. Cooley, Wulftange, et al. ²⁰²³). This strategy could be expanded to pathologies outside of oncology.

Therapeutic studies employing bubble cavitation have primarily been completed with MBs. However, recent evidence has shown that NBs can have a much more selective effect, sparing normal tissue outside the target zone (Nittayacharn et al. 2024). Transient blood–brain barrier opening, a potential avenue for cavitation effects, has previously been tested with NBs compared to commercially available MBs (Cheng et al. 2019). In their study, Cheng et al. reported improved consistency of harmonic emissions and acoustic control from NBs (Cheng et al. 2019). Further work can be done to identify the minimum dosage needed to elicit these therapeutic effects.

Overall, research related to NBs is underexplored. With improved understanding of these tiny agents is the potential to expand the applications of US contrast agents and their role in imaging diagnostics and US‐mediated therapeutics.

8. Conclusion

US imaging with NBs is an emerging and promising technique. Combining the principles of US imaging with the unique properties of NBs, recent work has established enhanced contrast and expanded imaging capabilities. CEUS is currently used worldwide with clinical indications in cardiology and radiology, and it continues to evolve and develop through innovative technological advancements. Due to the broad abilities of NBs, NB CEUS imaging continues to gain traction with new studies establishing their high echogenicity and superb deep tissue access. The aim of this review is to highlight the achievements in the field of nonlinear NB CEUS imaging including discussion on NB formulations, and their acoustic characteristics. The field of US imaging with NBs is still in its early stages, but it has shown great potential in preclinical research and animal studies. As technology advances and more research is conducted, this technique may find applications in various areas of medicine, including cancer detection and treatment, cardiovascular imaging, and drug delivery.

Author Contributions

Dana Wegierak: conceptualization (lead), data curation (lead), formal analysis (lead), investigation (lead), project administration (lead), writing – original draft (lead), writing – review and editing (lead). Pinunta Nittayacharn: writing – original draft (supporting), writing – review and editing (supporting). Michaela B. Cooley: writing – original draft (supporting), writing – review and editing (supporting). Felipe M. Berg: writing – original draft (supporting), writing – review and editing (supporting). Theresa Kosmides: writing – original draft (supporting), writing – review and editing (supporting). Dorian Durig: writing – original draft (supporting), writing – review and editing (supporting). Michael C. Kolios: supervision (supporting), writing – original draft (equal), writing – review and editing (equal). Agata A. Exner: resources (lead), supervision (lead), writing – original draft (equal), writing – review and editing (equal).

Conflicts of Interest

The authors declare no conflicts of interest.

Related WIREs Articles

Current status and prospects for microbubbles in ultrasound theranostics

Acoustically active liposome‐nanobubble complexes for enhanced ultrasonic imaging and ultrasound‐triggered drug delivery

Ultrasound imaging beyond the vasculature with new generation contrast agents

Multimodal micro, nano, and size conversion ultrasound agents for imaging and therapy

Breaking free from vascular confinement: Status and prospects for submicron ultrasound contrast agents

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.