Improving the Heating Efficiency of High Intensity Focused Ultrasound Ablation Through the Use of Phase Change Nanodroplets and Multifocus Sonication

Abstract

Thermal ablation by ultrasound is being explored as a local therapy for cancers of soft tissue, such as the liver or breast. One challenge for these treatments are off-target effects, including heating outside of the intended region or skin burns. Improvements in heating efficiency can mitigate these undesired outcomes, and here, we describe methods for using phase-shift nanodroplets (PSNDs) with multi-focus sonications to enhance volumetric ablation and ablation efficiency at constant powers while increasing the pre-focal temperature by less than 6°C. Multi-focus ablation with 4 foci performed the best and achieved a mean ablation volume of 120.2±22.4 mm₃ and ablation efficiency of 0.04 mm₃/J vs an ablation volume of 61.2±21.16mm₃ and ablation efficiency of 0.02 mm₃/J in single foci case. The combined use of PSNDs with multi-focal ultrasound presented here is a new approach to increasing ablation efficiency while reducing off-target effects and could be generally applied in various focused ultrasound thermal ablation treatments.

I. Introduction:

Locoregional therapy is the process of directly treating a diseased region of the body with the goal of improving prognosis. While surgical removal is the most common locoregional procedure, there is a constant push to make these procedures less invasive, which helps reduce morbidity, cost, and the need for hospitalization (Brem 2018). Locoregional treatment of cancers of soft tissues, such as the liver and breast, have seen significant innovation in the past decade. Hepatocellular carcinoma (HCC) is one of the most common malignancies of the liver (Stuver and Trichopoulos 2008) (Grandhi et al 2016), and although surgery is potentially curative treatment for HCC, it is contraindicated in cases where a single lesion <5cm in diameter exists or fewer than 3 lesions with diameters <3 cm (Ryder and Gastroenterology 2003). In cases where resection is not recommended, locoregional ablative therapies, such as radiofrequency ablation (RFA) or transcatheter arterial chemoembolization (TACE), can be effective to prolong life and manage pain (Inoue et al 2014, De Lope et al 2012, El-Serag 2012, Livraghi et al 2011, Nishikawa and Osaki 2013). In the breast, researchers have explored alternative treatments like cryoablation (Sabel et al 2004) where they create an ice-ball near the tumor to engulf the tumor using argon based cryoprobe. Other methods include radiofrequency ablation (Fornage et al 2007) and microwave ablation (Zhou et al 2012). RFA and TACE have established the efficacy of local ablative therapy, but they are invasive procedures, and an extracorporeal method that could achieve comparable outcomes would be desirable.

High Intensity Focused Ultrasound (HIFU) is increasingly been explored as an ablation option since it can focus sound waves of high pressures and intensity at a desired point to cause ablation non-invasively. In one clinical trial to treat liver and kidney tumors, (Middleton et al 2005), all liver tumors (100%) and a majority of kidney tumors (67%) were successfully ablated based on radiological or histological assessment after HIFU treatment. However, side effects in form of skin toxicity and blisters of < 1cm in the region of ablation and a low-grade fever in 4 out of 30 patients were observed. In a more recent hepatocellular carcinoma clinical trial by Zhang et al.,(Zhang et al 2009) the efficacy of HIFU in causing necrosis was investigated while assessing its effects on adjacent blood vessels. This study included ablation in 39 patients (with 42 tumors) with an average tumor size of 7.4±4.3 cm in its greatest dimension. Twenty-one of the 42 tumors were completely ablated while the rest were more than 50% ablated. The power levels used were between 160–250W at 0.8 MHz. Although most patients saw more than 50% tumor ablation, 9 patients experienced mild local pain. Among those 9 patients, 5 patients had mild skin burn, one patient had skin blisters and another patient had skin burns and blisters. Similar results were seen by Leslie et al., (Leslie et al 2012) where they treated 31 patients with primary and metastatic liver tumor with HIFU. 39% of patients received superficial skin burns and one patient observed with moderate skin burn. Similar outcomes are also seen when HIFU is used for breast cancer ablation. Breast cancer is one of the leading causes of death from cancer in women. To overcome these side effects, scientists used HIFU to ablate breast tissue in clinical trials (Gianfelice et al 2003, Merckel et al 2016, Wu et al 2003, Furusawa et al 2006). Some of these clinical trials report an ablation volume of around 8cm₃ at power levels between 80–271 W for 35–133 minutes. Similar to treatment trials with hepatocellular carcinoma, off-target heating was observed in breast cancer clinical trials as well.

Hence, despite the promise of HIFU technology for non-surgical ablation, off-target heating remains a major concern. The ability to reduce off-target heating while maintaining HIFU’s ability to provide non-invasive volumetric ablation would improve HIFU’s risk profile as a treatment for HCC as well as other diseases.

One approach to improving heating is to inject particles that accelerate heating due to ultrasound. Microbubbles can increase volumetric ablation because they can enhance acoustic absorption (Phillips et al 2013) and can decrease therapy duration(Kripfgans et al 2014). When sonicated with high enough pressure and appropriate frequencies, microbubbles oscillate and generate much more heat than in the absence of microbubbles. Microbubble oscillations induce local heating through viscous losses in the surrounding medium (Hilgenfeldt et al 2000), and can also produce local shock waves, which can aid in tissue ablation and cause mechanical stress at cellular level. Researchers have accelerated heating using Levovist, a clinically approved contrast agent with diameter less than 8 μm (Ignee et al 2016, Kudo 2003). Kaneko et al., (Kaneko et al 2005) reported a nearly double volumetric ablation in rabbit liver which was injected with 7mL of Levovist (371±104 mm₃) vs rabbit liver that was only injected with saline (166±71 mm₃). However, the use of microbubbles in a clinical setting to aid tumor ablation faces several challenges. Fundamentally, microbubbles tend to induce off-target heating, primarily in the near field (Phillips et al 2013). This is because microbubbles amplify thermal energy deposition regardless of their location. The potential of using microbubbles for enhancing focal heating is further limited since high concentrations of microbubbles can attenuate acoustic penetration into deeper tissues. Furthermore, microbubbles are too large extravasate from the vascular space, and remain localized within blood vessels, (Ferrara et al 2007, Villanueva and Wagner 2008), and finally, they have a short circulation half-life (Mullin et al 2011).

Prior data suggests that phase-shift nanodroplets (PSNDs) might serve as an improved alternative agent to microbubbles for enhancing HIFU thermal deposition. PSNDs are nanometer-sized liquid particles that are intravenously injected. They undergo phase shift and convert to microbubbles when sonicated above a threshold pressure. This threshold pressure can be tailored depending on the PSND composition (Sheeran and Dayton 2014). Because PSNDs only convert to microbubbles in regions of the pressure field above a threshold, they amplify heat only in the focal region, which can be used to reduce off-target heating. Moyer et al. observed a temperature rise of 130% and an ablation volume 30 times greater with PSNDs vs without PSNDs in the rat liver (Moyer et al 2015a) in response to HIFU treatment. Finally, the stability and circulation time of certain PSND formulations can be substantially longer than that of microbubbles (Sheeran et al 2015).

Volumetric ablation efficiency is a challenge for HIFU since the focal region is typically an ellipsoidal focus with diameters on the order of 1–3 mm. Clinical systems will typically steer a single focus in a circular or spiral pattern to generate a larger focal volume with larger spirals resulting in increased overall heating efficiency(Kim et al 2012, Partanen et al 2013). Another approach to improve heating is to use multi-focal heating pattern. Multi-focal heating patterns avoid high intensity levels that can develop due to high focal intensity levels when treating large tumors using ‘point by point sonication’ method. Ebbini et al (Ebbini and Cain 1989a) showed that using multi-focus treatment approach, one can precisely control the intensity levels at each of the control points in the treatment volume. This is especially useful for mild hyperthermia setting where using multiple concurrent foci can provide the benefit of reducing acoustic pressure near the focal region and providing better control over treatment volume (Dreher et al 2012, Chaplin and Caskey 2017). Dreher et al., (Dreher et al 2012) showed that multi-focus ablation resulted in a significant reduction (67%) in peak acoustic pressures in simulation and hydrophone measurements as compared to single focus. While multi-focal heating provides better spatial coverage, it also lowers overall energy efficiency of the region and concentrates more energy in regions away from the focal region. PSNDs pair well with multi-focal heating methods because they amplify thermal effects only in the focal region, keeping off-target regions safe. In this paper, we explore how pairing beam focusing patterns with PSNDs increase volumetric ablation and improve ablation efficiency.

Our overall goal is to increase the volumetric ablation while reducing the total energy delivered. We examined thermal effects of PSNDs on the lesion size in phantoms during HIFU ablation using magnetic resonance (MR) thermometry in 7T MRI. Using the temperatures obtained via MR Thermometry, the values of volumetric ablation that reached CEM 240 at 43°C, i.e the lethal thermal dose, was calculated as it has been done in previous studies (Kim et al 2012). Lastly, ablation efficiency (mm₃/J) was calculated, and cases of single focus ablation was compared with multi-focus ablation with and without PSNDs.

II. Materials and Methods

Lipid Solution

The procedure for creating a lipid solution for nanodroplets (PSNDs) was originally published in Moyer et al.(Moyer et al 2015b). Two lipids were combined in a 1:9 ratio (both from Avanti Polar Lipids, Alabaster, AL, USA) to create the shell of the PSNDs: 1, 2-distearoyl-sn-glycero-3-phosphoethanolamine-N-methoxy (polyethylene-glycol)-2000 (DSPE-PEG), and 1, 2-distearoyl-sn-glycero-3-phosphocholine (DSPC), respectively. Lipids were emulsified in an 1.0 mg/mL aqueous solution of 15% (w/v) propylene glycol, 5% (w/v) glycerol, and 80% (w/v) phosphate-buffered saline (PBS). Once prepared, the lipid solution was portioned into 1.5mL aliquots and stored in 3mL glass vials. Vials are sealed and stored at 4°C.

Nanodroplet Protocol and Size Distribution

When ready for use, lipid solutions were removed from storage, and the headspace of the vial was replaced with desired perfluorocarbon (PFC) gas. In this study, a 1:1 mixture of dodecafluoropentane (DDFP, C5F12, boiling temperature= 28°C), and decafluorobutane (DFB, C4F10, boiling temperature= −2°C). The PFC mixture was obtained by filling a syringe with 60mL of DFB (Fluoromed, Round Rock, TX, USA), which was contained in a gas canister set to 6 psi. The syringe was placed in a −20°C freezer, and the DFB was condensed to a liquid. Liquid DFB was mixed with 0.5mL of DDFP (Exfluor Research Corporation, Round Rock, TX, USA), and the liquid solution was vaporized back into a gas. The DDFP-DFB gas mixture was then connected to the lipid solution vials and filled the vacuumed headspace. The PFC containing lipid solution was shaken into a microbubble solution for 45 sec (VIALMIX® Shaker, Lantheus Medical Imaging, N. Billerica, MA, USA). The vials were placed in a −11°C ethanol-water bath for 2 min, and then, 14 psi of nitrogen gas was pumped into the headspace to pressurize the microbubbles into PSNDs. The vials continued to be swirled in the cold bath for 30 sec, before the nitrogen line was removed, and vials were placed on dry ice for shipping. The vials of PSNDs were shipped overnight to Vanderbilt University on dry ice.

Size distributions of the PSNDs were collected using a NanoSight NS500 (Malvern Instruments, Westborough, MA, USA). A 2:998 dilution of the PSNDs was used to characterize the particles, by adding 2 μL of stock PSNDs to 998 μL of 20nm-filtered HPLC-grade water. In Figure (1), a total of 3 vials was measured to create the size distribution presented.

Figure (1):

This figure presents the DDFP-DFB nanodroplet size distribution

Tissue-mimicking phantoms

Acrylamide-albumin tissue mimicking phantoms were created from an 8:7:5 mixture of aqueous acrylamide (bis-acrylamide to acrylamide ratio of 1:29), liquid egg-white, and deionized water, respectively. This is similar to the process described previously (Takegami et al 2004). These phantoms have acoustic properties very similar to that of tissues. Their density is 1.0g/cm3, sound speed of 1540 m/s, and acoustic attenuation of 0.4 dB/cm at 2 MHz. When heated to 60°C or above, the egg whites start to permanently denature and create an opaque lesion. At temperatures below 60°C, they are nearly transparent. These phantoms have nanodroplets of the concentration 0.8ul/1ml of phantom material. The nanodroplets were dispersed to acrylamide solution before polymerization via the addition of 0.5% vol/vol of 10% ammonium persulfate and 0.2% vol/vol of tetramethylethylenediamine. This solution was then poured into a cylindrical mold of 40mL. We used these phantoms within 24 hours of their polymerization.

High intensity Focused Ultrasound

HIFU was delivered using a spherically focused, 128 elements, 650 kHz phased array (Image-Guided Therapy, France). This array is randomized with elements 6.6mm in diameter. The radius of curvature and opening diameter is 7.2 cm and 10.3 cm respectively (Chaplin et al 2018). A continuous wave HIFU was applied using this transducer to ablate phantoms and was an average power of 80 W. The power reported is based on electrical measurements of voltage and current delivered to the transducer. It is derived from measuring voltage (V), current (I) and phase (theta) between voltage and current. During a sonication with our transducer, the RMS power is P=IV*cos(theta). The powers reported here are calculated from averaging multiple reading dissipated in our transducer by our system. Pressures at each power level at natural focus were initially recorded by Fiber Optic Hydrophone (Precision Acoustics, Dorchester, UK). The pressure (PNP and PPP) recorded from Fiber Optic Hydrophone ranged from 0.5–8.1 MPa and 0.5–13MPa respectively. We then performed pressure calibration at each power until 2.2W via a needle hydroplane (ONDA HNA-0400, Sunnyvale, CA) in a degassed deionized water bath. These values were subsequently used for extrapolation for pressures at higher powers based on fiber optic hydrophone measurements. Extrapolation of values using needle hydrophone showed a PNP of 4.2 MPa and this value matched the pressure value recorded by Fiber Optic Hydrophone at that power. The experimental set up resembled Figure (3). Different cases of activation and heating are presented in Figure (2) and elaborated more in section below. A burst of 20 pulses, each with duration of 0.09s, were used for PSND activation. The exact particle activation sequence varied between cases to ensure that all particles in the region were activated prior to heating pulses: a) for single focus, 1 burst was applied with 100W power; b) for 2 foci a single multi-focal burst was transmitted at 200 W; c) for 4 foci ablation two multifocal bursts at 200 W were applied to activate 4 foci, and d) for the 5-mm diameter circle, 8 bursts at 100 W were used to activate each of the 8 points. Based on our extrapolated hydrophone estimates, these pulses ensured that each focal point in single or multi-focal cases received a focal pressure and duration known to convert particles (PNP 4.2MPa - 4.4 MPa, PPP 4.5MPa-4.6MPa ), confirmed by B-mode images (Figure 4). The activation pulses are the only difference in energy applied between each study group. Conversion of nanodroplets was followed by a 48 s heating pulse at 80W.

Figure (3).

(a) This figure is of side view of our experimental set up which consists of water bath and 5% agar phantom holder in that water bath. This agar phantom holder has a slot for our egg-white acrylamide phantom. The transducer is attached to the water bath and the agar phantom holder and egg white phantom is placed right in front of the transducer at a distance where the beam from the transducer converges to form the focus (red ellipse). This set up of transducer, water bath and our phantoms is then moved into the 7T MR bore encased in single channel NOVA coil. (b) A zoomed in representation of total number of slices that were acquired in our phantom and how the focus aligned with the slices. Each slice was evenly spaced apart by 2mm. Additionally, it shows location of slice that was chosen as to evaluate surface heating with respect to slice that saw most heating and was chosen as slice for focal heating. Slice 2 was 8mm away from the focus. It was important to choose slice inside the phantom for surface heating in order to provide accurate measurements of heating for conditions that use PSNDs.

Figure (2) –

The top row presents how the focal points look in 2D space. The bottom row presents the Rayleigh Sommerfeld pressure simulation at those foci. Following are the specific cases:(a) & (e) Case 1 where the activation and heating happened at the natural focus.(b) and (f) Case 2 where activation and ablation happened point wise on an 8-points 5mm diameter circle. Each of the 8 points were first activated and was followed by heating. (c) and (g) Case 3 where two points were activated at the same time and followed by heating. Heating at both points took place at the same time. (d) and (h) Case 4 where 4 foci was activated and heated. In this particular case 1^st 2 foci were activated and then heated. Once activation and heating of 1^st 2 foci was done, same happened to 2^nd 2 foci.

Figure (4):

Contrast images acquired using imaging US after activating and heating phantoms. The red dots represent location of the focus in each case. Contrast imaging helps in identification of microbubbles in a medium as they give signals out at higher harmonics. We used this imaging technique to ensure conversion of our nanodroplets to microbubbles. (a) Contrast imaging obtained after activating and heating natural focus located at (0,0) in the image. (b) Contrast image obtained after activating and heating 8 point 5mm diameter circle. 7 out of 8 points saw heating in this case. (c) Contrast image obtained after activating and heating 2 points 5mm away from each other. Both points saw activation and heating. (d) Contrast image obtained after activating and heating 4 points – 2 points 5mm apart were heated and activated at a time. Once first 2 points were activated and heated, next 2 points went through the same regime. In this particular case, 3 out of 4 points saw activation and heating.

Groups of phantoms for treatment

Phantoms were made and then distributed in different groups. The groups consisted of phantoms that would undergo one of the following treatments presented in Table (1). A total of 32 experiments were done. Total energy delivered in each case was 3840 J and average power was 80 W. All phantoms were heated for 48 seconds with a near 100% duty cycle. Multiple foci was generated using method described by Ebbini et al (Ebbini and Cain 1989b) using our 128 random phased array spherical cap(Chaplin et al 2018). Figure (2) presents all our cases. The top row represents the foci in 2D space and bottom row represents the pressure map for those points. The case 2 is similar to how single focus sweep sonications were done in prior studies (Dreher et al 2012).

Table (1):

Egg-white acrylamide phantoms with and without PSNDs were subjected to 4 different treatment options above. There were in all of 32 observations, 8 in each group. In that, 4 observations in each group was carried out on phantoms with PSNDs and 4 observations were carried out on phantoms without PSNDs.

	Natural focusy	2 foci	4 foci	5 mm diameter circle
Activation Power (in W)	100	200	200	100
Heating power (in W)	80	80	80	80
Total time of sonication (in s)	48	48	48	48
Total energy delivered (in J)	3840	3840	3840	3840
Number of observations	8	8	8	8

Generation of Multiple foci

Single focus sonications were performed at the transducer’s natural focus. In order to generate multiple foci, methods outlined in previous works(Ebbini and Cain 1989b). The method proposed in this paper uses phasing method for synthesis of multiple focus. This section covers the derivation of this method. The complex pressure at a point in acoustic field can be represented by:

$$p(r)=\frac{j\rho ck}{2\pi }{\int }_{S^{\prime }}u\left(r^{\prime }\right)\frac{e^{\left(-jk\left|r-r^{\prime }\right|\right)}}{\left|r-r^{\prime }\right|}d{S}^{\prime }$$ (1)

where $j=\sqrt{-1}$, ρ is the density, c is the speed of sound, k is the wavenumber, S′ is the surface of the source, u is the normal velocity of the source surface, and r and r′ are the observation and source points, respectively.

If a phased array transducer has N number of elements with surface area $S_n^{\prime }$ located at $r^{\prime }=r_n^{\prime }$ with particle velocity u_n, pressure at M control points at position r = r_m where m = 1,2,… M can be written as

$$p\left(r_m\right)=\frac{j\rho ck}{2\pi }\sum _{n=1}^N{u}_n{\int }_{S^{\prime }}\frac{e^{(-jk|r_m-r_n\prime |)}}{\left|r_m-r_n\prime \right|}d{S}_n^{\prime }$$ (2)

where u_n is removed from the integral by assuming the transducer element has constant particle velocity across its surface. If we let

$$H(m,n)=\frac{j\rho ck}{2\pi }{\int }_{S^{\prime }}^{■}\frac{e^{\left(-jk\left|r_m-r_n\prime \right|\right)}}{\left|r_m-r_n\prime \right|}d{S_n^{\prime }}^{■}$$ (3)

then we can represent equation 1 in matrix form where:

$$Hu=p$$ (4)

where u = [u₁,u₂,… u_N ]^t and p = [p(r₁),p(r₂),… p(r_M )]^t

One can use equation 4 can help solve for amplitude and phase of N transducer elements that can generate desired pressure at M control points. When M<N, H has full rank and minimum-norm solution of the matrix in equation iv can be solved by:

$$\widehat{u}=H^{*t}{\left(H{H}^{*t}\right)}^{-1}p$$

where H*^t is the conjugate transpose of H. The resulting $\widehat{u}$ will be a matrix of size [Nx1] where each complex number corresponds to amplitude and phase of the n_th element.

The multiple foci patterns used in study were dictated by the frequency (650 kHz), focal size of our transducer (full width half maximum was 9.3mm and 2.2mm in the axial and lateral directions), and pressure threshold for PSNDs conversion (Chaplin et al 2018). We sought to perform patterns that closely match the studies performed by transducers at frequencies of 1.2 MHz with focal spot size of 1 × 1 × 8 mm₃ (Mougenot et al 2011, Dreher et al 2012). The 5-mm diameter circle is comparable to ablation cell sizes used in clinical systems like Sonalleve (Mougenot et al 2011, Partanen et al 2013). The choice of having points 5mm away from each other ensured that points did not overlap and at the same time generated enough pressures to convert PSNDs to MBs.

Recording of Onset of Vaporization –

The onset of vaporization followed by heating induced by these nanodroplets was observed using e17–4 probe of Phillips EPIQ7 ultrasound scanner machine in their contrast mode. The image obtained was in DICOM format and MATLAB R2018a was used to read the data. Acoustic contrast images have the capability of confirming the vaporization of nanodroplets as this imaging technique identifies regions that give out signals at higher harmonics and help identify microbubbles in those regions.

MR-Thermometry –

MR thermometry was employed to measure the volume of heat deposition and maximum rise in temperature. MRI was performed using human Philips Achieva 7 Tesla MRI (Best, Netherlands) using a NOVA Medical (Wilmington, MA, USA) birdcage for both transmit and receive. Sonications were performed inside the scanner using our MR-compatible HIFU device. Albumin-acrylamide gel phantoms were insonated using continuous wave HIFU at 650 kHz. The first pulse was the activation pulse at 100 W for single focus and at 200 W for multi-focus. After the first pulse, the phantoms were heated at lower power to deliver a total energy of 3840 J in a span of 48 seconds. The average power was consistent across phantoms with standard deviation of ± 3%. Temperature rise and volumetric heating was monitored using 3D sagittal slices that were 2mm apart. All images were acquired using 2D multi-slice spoiled gradient echo with following parameters

TE (echo time) = 10ms

TR (repetition time) = 25 ms

Field of view = 120mm x 26mm x 120mm

In plane resolution – 2mm x 2mm

Flip angle = 10 degrees

Slice thickness – 2mm

Scan time – 120 seconds

For each application of ultrasound sonication, a total of 13 slices were acquired. Sonication was started 21 seconds into imaging and the heating continued for next 48 seconds. The total scan time was 2 minutes. Temperature maps were reconstructed using MR images by employing proton resonance frequency shift method(Ishihara et al 1995, Hindman 2005). The temperature change was calculated by computing the phase difference between baseline images and images that proceeded right after heating was on assuming α = 0.01ppm/°C and field strength of 7T. The thermometry and imaging plane are presented in Figure (3). The data acquired from 7T MRI was processed off-line in MATLAB R2018a. The heating pixels were unwrapped using phaseUnwrap2D algorithm(Herráez et al 2002). After unwrapping, the temperature rise in each slice was calculated using the phase difference in the images. The maximum ΔT observed in focus of each phantom with PSND and the maximum ΔT observed in slice 8mm away from the focus of that phantom was then calculated to quantify focal and pre-focal heating. After evaluating the rise in temperature, volume that reached 240 cumulative number of equivalent minutes at 43 °C (240 CEM 43 °C) was calculated by setting the base temperature to 37 °C. The CEM 43 °C was calculated using the formula:

$$CE{M}_{43}=\sum _{i=1}^N{\left[R_{CEM}\right]}^{\left(43-T_i\right)}{t}_i$$ (5)

where ti is the i-th time interval, R is related to the temperature dependence of the rate of cell death (R(T< 43 °C)=1/4, R(T>43 °C)=1/2) and T_i is the average temperature during time interval t_i. CEM₄₃ values were used for creating isosurfaces (figure 6(c)) for volumes greater than 240 CEM 43 °C to study the shape of lesion formed.

Figure (6):

(a) This graph represents the mean volume that reached at least 240 CEM 43 °C in phantoms with PSNDs. This was calculated by calculating the temperature in CEM 43 and thresholding all voxels in MR Thermometry slices that were greater than 240 CEM 43 °C. (b) Tabulated form of mean volume in each case. c) These images represent isosurface of volumetric ablation. It gives the spatial distribution of the ablated volume and shows single, 2 foci and 4 foci had contiguous lesions but circle does not. This is only for one phantom with PSNDs in each case but this scenario was also true for other phantoms with PSNDs in all cases.

III. Results

Nanodroplet Size Distribution

We found that the NDs formulated with a DDFP-DFB core had an average mean diameter ± standard deviation of 219 ± 67 nm (N=3). The average concentration for the size distribution was 2.9×10₁₁ ± 6.3×10₁₀.

B-Mode contrast image of phantoms with PSNDs:

The results of our B-Mode imaging demonstrate that phantoms with nanodroplets can undergo phase transition at 4.4 MPa followed by heating pulse power matched at 80W (Figure 4). The size of focal zone at the natural focus is consistent with lateral width of the beam at focus, which is ~2mm (Figure 4(a)). In the case of circle (Figure 4(b)), activation can be seen 6 out of 8 of the desired points. Similarly, in the cases of 2 foci (Figure 4(c)) and 4 foci (Figure 4(d)) activation can be seen at desired locations. In order to get a spatial distribution of temperature values, we performed MR thermometry (discussed in later sections) which revealed that heating at an accelerated rate consistent with activation was observed in all desired locations for multiple focus.

On target vs Off target heating

We acquired MR thermometry images using Philips 7T MRI to quantify the rise in temperature at the focus (focal heating) and to quantify rise in the temperature at pre-focal surface of the phantom (pre-focal heating). The maximum rise in temperature at the desired focus (Figure 5(a)) in phantoms bearing PSNDs is comparable across cases where a rise of 32.5 °C, 30.4 °C, 32.5°C, and 32.5 °C was seen in single focus, circular pattern, 2 foci and 4 foci cases respectively. The maximum temperature rise in the pre-focal region was 3.1 °C, 3.5 °C, 3.1 °C, and 3.0 °C for single focus, circular pattern, 2 foci and 4 foci cases respectively as seen in Figure 5(b).

Figure (5):

(a) Graph that shows maximum rise in temperature at the focus and (b) at the surface. The surface was 8mm away from the slice that saw the most heating. The values in this graph is comparable across all 4 cases. 4 foci saw the greatest focal heating, only slightly better (0.04°C greater) than 2 foci heating. But 2 foci performed the best when the maximum rise in surface heating was seen. It showed the lowest surface heating of 3.123 °C

Phantoms without PSNDs experienced less heating. Maximum temperatures rose to 7.7°C, 6.8°C, 6.4°C, 7.6°C in single focus, circular pattern, 2 foci and 4 foci cases respectively (Figure 5(a)). The maximum temperature rise in pre-focal region was 3.4 °C, 4.6 °C, 5.9°C, and 4.2 °C for single focus, circular pattern, 2 foci and 4 foci cases respectively (Figure 5(b)).

Volumetric ablation calculation

After obtaining the change in temperature values and adding 37 °C to those temperature values, we calculated 240 CEM 43 °C using equation 5. We used voxels that had reached a value greater than 240 CEM 43 °C to evaluate the final volumetric ablation (Figure 6(a)). The volumetric ablation achieved by 2 foci and 4 foci were consistent and greater than volumetric ablation achieved by single focus and circle pattern as shown by the rate of change of mean volumetric ablation (Figure 6(a)). The average volumetric ablation increased during the ultrasound period in all heated phantoms bearing PSNDs (Figure 6(a)). Representative volumetric isosurfaces of voxels with temperature rises greater than 240 CEM 43°C show how 4 foci, 2 foci and single focus develop contiguous lesions, while circular foci often had gaps between foci (Figure 6(c)). There was a significant difference (p < 0.05) between volumetric ablation of single focus and 4 foci based on two-sided Wilcoxon rank sum test; however, there was no significant difference between volumetric ablation of circle and single focus (p >0.05) and between single focus and 2 foci (p=0.0571). There was also no significant difference between two multi-focus cases; however, there was a significant difference in volumetric ablation between multi focus (both 2 foci and 4 foci) and ablation in circle pattern (p < 0.05). Figure 6(b) shows mean 240 CEM at 43°C over 120 seconds. In this case, 4 foci performed the best and circle performed the poorest.

Evaluating ablation efficiency

After obtaining the volumetric ablation in all cases, we calculated ablation efficiency (Figure 7) where the volumetric ablation (in mm₃) in each case was divided by the total energy put in (in J). Again, 2 foci and 4 foci saw the greatest ablation efficiency values of up to 0.04 mm₃/J for each. Single foci and circle patterned foci saw maximum ablation efficiencies of 0.023 mm₃/J and 0.02mm₃/J respectively.

Figure 7:

Graph representing maximum ablation efficiency. This was calculated by dividing ablation volume that reached greater than 240 CEM 43 °C with the total energy deposited to each phantom. Value for each case is presented here. Multi-foci performed the best here.

IV. Discussion

We used HIFU in conjunction with PSNDs to improve the efficiency of HIFU thermal ablation. We developed a method where increased volumetric ablation for a fixed amount of energy could be delivered and tested it on phantoms with acoustic properties similar to tissues. The increased volumetric efficiency when combining multi-focal sonications and PSNDs could improve volumetric ablation in clinical cases while decreasing off-target heating, overcoming two of main challenges facing thermally ablative HIFU procedures.

Heating restricted to focal region with PSNDs

Phase shift nanodroplets were activated in the shape of chosen beam patterns. The heating was contained within the desired regions and no off-target regions underwent any ablation as seen in Figure 6(c). The natural focus activation pattern spans the 2-mm lateral beam width of the focus, while multi-focal and circularly steered cases have activation patterns that span the expected ranges (Figure 6(c)). Our sonication procedure resulted in a mean rise of only 6°C at the pre-focal region of the phantom which implies that ablation using PSNDs only results in on-target heating and minimizes any off-target heating effects and may potentially reduce the severity of incidences like skin burns. In the absence of PSNDs, the maximum focal heating observed was less than 7°C compared to >30°C in the presence of PSNDs. Similar to prior studies, heating was restricted to the focal region with very little heating occurring pre-focally (Figure 5 (b))(Moyer et al 2015b).

Increased volumetric ablation and ablation efficiencies observed in MF-PSND when compared with SF PSNDs at matched power levels.

Multi-focus heating at a constant energy and power generated an increased volumetric ablation when compared to single focus ablation at transducer’s natural focus. The heated volume reaching CEM43 of 240 based on MR thermometry was twice as large in the multifocal cases compared to natural focus or circularly steered (Figure 6). Four foci ablation achieved the highest volume of 120.2±25.8 mm₃ which was followed by 2 foci ablation which achieved a thermal ablation volume of 111.7±29.03 mm₃, all at 80W.

Heated volumes observed in MF-PSND were contiguous when compared to single focus circle case.

Heating was observed to be contiguous for multi-focus cases with no gaps present in between the multiple foci as demonstrated in Figure 6(c). This was not true for the sequentially scanned circular pattern, which showed gaps that did not reach CEM43 of 240 after sonications. Thermal distribution with multiple foci is higher than the single point (Wu and Sherar 2002). In the single focus case, volumetric heating beyond the focal region only occurs due to diffusion of heat. The circular scan distributed heat over a volume comparable to the multifocal patterns, but the heated volume did not reach CEM43 of 240 despite having matched power levels. This could be due to shorter duration of heating pulses per focus. For the case of ablation performed in the circle, each point on the circle only received heating for 6 seconds vs 48 seconds on other cases. Hence, longer duration of heating pulses may have caused diffusion of heat between adjacent nanoparticles, thereby causing them to convert as heating can also aide in phase shifting the nanodroplets (Zhao et al 2016). Our maximum observed efficiency of 0.04 mm₃/J in 2 foci and 4 foci ablation was almost twice than what was observed in single focus case. When these results were compared with 4mm circular trajectory HIFU ablation performed in clinical trials by Kim et al, they had an ablation efficiency of 0.06±0.12 mm₃/J which was higher than the ablation efficiency observed in our study. While the higher ablation efficiency in their study could be attributed to the 1.2 MHz center frequency, further analysis that includes frequency dependent attenuation between the transducer face and target would be required to quantitatively compare our PSND study to the in vivo study.

Study Limitations

Some limitations should be considered when interpreting our study results. Our study was aimed at improving ablation efficiency at constant power. This was achieved by combining multi-focus sonication sequences with PSNDs. These sonication patterns were chosen to closely match patterns in previous clinical trials but would likely benefit from optimization via simulation. Rayleigh-Sommerfeld integral simulations in conjunction with bioheat equation during electronic steering could have resulted in a better treatment plan (Mahoney et al 2001, Li et al 2011, Wu and Sherar 2002, Fan et al 2013) as it would have given us pressure fields and heat deposition estimation at those pressure fields. PSNDs change the acoustic absorption in a spatially specific manner and including this phenomenon is beyond the scope of the present study. To facilitate matched comparisons we did not implement any feedback control in our study, but feedback control could likely be optimized with the methods presented as in prior studies (Poorman et al 2016, Kim et al 2012, Voogt et al 2012). Our finding that multifocal HIFU coupled with nanodroplets has increased volumetric ablation efficiency is general and should be function synergistically with prior optimizations. Volumetric ablation at lower powers than those currently used in clinical trials could benefit many therapeutic ultrasound procedures, leading to decreased time when treating large regions and an increased safety margins due to the lower energy required.

Conclusion

This study successfully showed that PSNDs and multi-focus sonication can be used to increase focal ablation volume while avoiding any off-target heating when compared to performing volumetric ablation at single focus. This approach could substantially improve the clinical translatability of HIFU ablation by reducing treatment time and reducing side effects.