ULTRASTRUCTURAL ANALYSIS OF VOLUMETRIC HISTOTRIPSY BIOEFFECTS IN LARGE HUMAN HEMATOMAS

Abstract

Large volume soft tissue hematomas are a serious clinical problem, which, if untreated, can have severe consequences. Current treatments are associated with significant pain and discomfort. It has been demonstrated in an in vitro bovine hematoma model that pulsed high intensity focused ultrasound (HIFU) ablation termed histotripsy can be used to rapidly and non-invasively liquefy the hematoma through localized bubble activity, enabling fine-needle aspiration. The goals of this study were to evaluate the efficiency and speed of volumetric histotripsy liquefaction using a large in vitro human hematoma model. Large human hematoma phantoms (85 cc) were formed by recalcifying blood anticoagulated with citrate phosphate dextrose (CPD) - saline-adenine-glucose-mannitol solution (SAGM). Typical boiling histotripsy (BH) pulses (10-ms or 2-ms) or hybrid histotripsy pulses using higher amplitude and shorter pulses (0.4-ms) were delivered at 1% duty cycle while continuously translating the HIFU focus location. Histotripsy exposures were performed under ultrasound guidance with a 1.5 MHz transducer (8 cm aperture, F# = 0.75). The volume of liquefied lesions was determined by ultrasound imaging and gross inspection. Untreated hematoma samples and samples of the liquefied lesions aspirated using a fine needle were analyzed cytologically and ultrastructurally with scanning electron microscopy (SEM). All exposures resulted in uniform liquid-filled voids with sharp edges; liquefaction speed was higher for exposures with shorter pulses and higher shock amplitudes at the focus (up to 0.32, 0.68, and 2.62 mL/min for 10, 2, and 0.4 ms pulses, respectively). Cytological and ultrastructural observations revealed completely homogenized blood cells and fibrin fragments in the lysate. Most of the fibrin fragments were less than 20 μm in length, but a number of fragments were up to 150 μm. The lysate with residual debris of that size would potentially be amenable to fine-needle aspiration without risk for needle clogging in clinical implementation.

INTRODUCTION

Intra-abdominal, retroperitoneal, and intramuscular hematomas are a serious clinical problem that can occur as a result of trauma, surgical intervention, bleeding disorders, blood cancer, and use of anticoagulant medications and blood thinners (Bovonratwet et al. 2019; Chung 2016; Dohan et al. 2015; Elhammady et al. 2011). Large hematomas can reach liters in volume, and depending on location in the body, can cause significant pressure on the surrounding tissues leading to serious complications, including organ failure, infection, loss of limb functionality, and amputation (Broderick et al. 1993; Chung 2016; Garner et al. 2014). Currently, there are no effective treatments for the immediate resolution of large hematomas. Standard of care includes drainage or surgical removal. The former is ineffective for acute clots due to their gelatinous nature, and the latter is associated with a high risk of recurrence, infection, and mortality (Li et al. 2020; Weiss et al. 2015). An alternative conservative treatment involves the gradual resolution of the hematoma over a long period, which is inapplicable in some cases due to rapidly emerging complications and pain (Smith et al. 2006). Moreover, certain types of hematomas tend to grow in size rather than resorb (Angster and Da Costa 2018).

Histotripsy, a technique utilizing nonlinear propagation of high intensity focused ultrasound (HIFU) pulses and shock formation at the focus, was recently proposed as a rapid and minimally invasive hematoma treatment under either ultrasound- or MR-guidance (Gerhardson et al. 2017; Khokhlova et al. 2016a; Li et al. 2020; Ponomarchuk et al. 2019). There are two main types of histotripsy: cavitation cloud histotripsy (CH) (Parsons et al. 2006) and boiling histotripsy (BH) (Khokhlova et al. 2015). CH utilizes a sequence of microsecond-long HIFU pulses with high peak negative focal pressure that generates bubble clouds at the focus; their activity mechanically homogenizes tissue (Lin et al. 2014b; Parsons et al. 2006). BH uses longer (milliseconds-long) pulses with lower focal pressures that initiate highly localized boiling within each pulse as a result of tissue heating by shock fronts that develop in the acoustic waveform at the focus due to nonlinear propagation effects (Khokhlova et al. 2011). Interaction between the remaining incoming shocks of the BH pulse with the expanding mm-sized vapor cavity results in tissue liquefaction (Khokhlova et al. 2011; Simon et al. 2012; Pahk et al. 2019; Pahk et al. 2021). Both histotripsy approaches create purely mechanical lesions in biological tissue with negligible thermal damage when a low duty cycle (DC) of about 1% is used. The efficacy of histotripsy for liquefaction of coagulated blood has been demonstrated in vitro both for vascular clots (Zhang et al. 2016) and large-volume hematomas (Gerhardson et al. 2017; Khokhlova et al. 2016a; Li et al. 2020).

Single lesions produced by BH and CH in large hematomas with the same ultrasound frequency and treatment time differ in shape and size. Single lesions produced by BH typically are “tadpole”-shaped, having a wider, round proximal region (“head”) and a narrow distal region (“tail”) that merge at the HIFU focus. The mechanisms responsible for such shape are still under discussion. Some studies suggest the incoming shock waves generate micro fountains at the pressure release surface of the boiling bubbles, thus creating the “head” of the lesion through acoustic atomization of the prefocal tissue. The “tail” of the lesion is then generated by mechanical “chiseling” of tissue by acoustic streaming of the debris ejected by the micro-fountains (Simon et al. 2012). Other studies suggest an additional mechanism for lesion formation; the lesion growth starts with the formation of the “tail” containing boiling bubbles which act as a pressure release surface for the incident waves resulting in the generation of cavitation clouds proximal to the focus, thus producing the “head” of the lesion (Pahk et al. 2019; Pahk et al. 2021). Both proposed mechanisms are likely to be involved in BH-induced tissue liquefaction, subsurface cavitation being an important part of the atomization process, and relative contributions of streaming and boiling bubble expansion to the “tail” formation depending on tissue mechanical properties. The relative roles of these mechanisms may also be heavily dependent on the HIFU frequency, even within a relatively narrow 1–2 MHz range (Khokhlova et al. 2017).

The lesion volume is mainly contained in the lesion’s “head”, and the focus scanning strategies in volumetric BH ablations usually aim to merge the “heads” of individual lesions, leaving the “tails” separate (Khokhlova et al. 2016a). BH-induced voids are larger in volume than those created by delivering CH-pulses which are more ellipsoidal (Khokhlova et al. 2016a). Recently, a “hybrid” histotripsy approach that uses sub-millisecond pulses (intermediate between BH and CH) was proposed and successfully tested in ex vivo porcine liver, kidney, and cardiac muscle (Eranki et al. 2018). The mechanisms involved in this approach are currently not fully understood. One hypothesis is that the use of longer HIFU pulses than CH heats the tissue, reducing the cavitation threshold leading to the initiation of cavitation cloud activity. Simultaneously, tissue heating at the focus accumulates and may initiate boiling after several pulses. Thus, both mechanisms may be involved in this hybrid approach. Single lesions generated by hybrid histotripsy are of similar shape but larger than the CH lesions and smaller than the BH lesions.

Volumetric histotripsy lesions are typically produced by generating lesions over a discrete spatial grid with specific spacing in transverse and axial directions allowing them to merge into a volumetric void (Gerhardson et al. 2017; Khokhlova et al. 2016a; Khokhlova et al. 2019; Li et al. 2020). It has been demonstrated that using tighter spacing with fewer pulses per sonication point results in more rapid volumetric liquefaction (Bawiec et al. 2020; Gerhardson et al. 2017). This study aimed to improve the liquefaction rate further and potentially provide more uniform liquefaction by using continuous HIFU focus translation within the targeted volume of hematoma. A secondary goal was to compare the efficiencies of the BH and hybrid regimes and the regularity of lesion borders.

The hematoma lysate generated with histotripsy should be amenable to fine needle drainage when considering the clinical treatment of hematomas. Thus, the residual fragments within the lysate have to be small enough not to clog the needle. In prior studies, the size distribution of the residual debris was measured by a Coulter Counter (Khokhlova et al. 2016a; Li et al. 2020). However, the upper limit of the measurement system in that studies was 70 μm which was determined by filtering the lysate before measurement to avoid clogging the system. One way to determine the sizes of larger debris fragments is through scanning electron microscopy (SEM). SEM of blood clot samples has been used extensively in studies of fibrinolysis (Weisel et al. 2008), the influence of RBCs’ presence on clot structure (Gersh et al. 2009), thrombolysis (Bester et al. 2018; Janis et al. 2002; Sutton et al. 2013) and other ultrastructural examination of clots (Pretorius et al. 2011). Here we suggest the use of SEM analysis for sizing of post-histotripsy fragments in liquefied lysate.

The overall goal of this study was to investigate and compare the efficiency and quality of BH and hybrid histotripsy in the volumetric liquefaction of large human hematoma phantoms under continuous HIFU focus translation. The measures of efficiency included the rate of liquefaction, the structure of the lesion borders, and the degree of hematoma disintegration, as evaluated by light microscopy and SEM.

MATERIALS AND METHODS

Human hematoma model

After obtaining informed consent, fresh human blood was collected from healthy volunteers through sterile venipuncture according to the guidelines and after approval of the review board of the Dmitry Rogachev National Research Center (Moscow, Russia). Citrate phosphate dextrose solution (CPD) was used as an anticoagulant, and saline-adenine-glucose-mannitol solution (SAGM) was used to provide improved preservation of red blood cells (MacoPharma S.A., France). The anticoagulated whole blood (450 ml) was stored in the refrigerator (+4°C) for one day. One day after blood collection, 85 ml of anticoagulated blood was poured into a rectangular plastic mold and degassed in a vacuum desiccator at 0.1 bar residual pressure for 1 hour. After degassing, the blood was recalcified with CaCl₂ solution (DalHimFarm, Russia) to a final concentration of 25 mM/L at 37°C for approximately 30 minutes till complete coagulation occurred (Khokhlova et al. 2020). Before the histotripsy exposure, the hematoma samples were embedded into a 6% agar gel (Kotanyi, Austria), which served as an acoustically transparent container for the samples as shown in Figure 1a.

FIGURE 1.

(а) Human hematoma phantom in agar gel. (b – c) A schematic (b) and a photograph (c) of the experimental setup for histotripsy liquefaction of human hematoma phantoms. B-mode ultrasound imaging was performed with Verasonics V1 imaging system and ATL L7–4 probe.

Histotripsy experimental arrangement

For histotripsy exposures, the hematoma phantom in the agar gel was transferred into a plastic holder, attached to a 3-D positioning system (Precision Acoustics UMS3, Dorset, UK) and immersed in degassed water (Precision Acoustics Water Treatment System, Dorset, UK) (Fig. 1,b – c). The treatment was performed 3 – 4 hours after the start of coagulation using a custom-built 1.5 MHz 12-element spherically focused transducer of 80 mm aperture, 60 mm focal length, and a 24 mm-diameter central opening (F# = 0.75) powered by a custom-built amplifier (Maxwell et al. 2017). Here F# or F-number is defined as the ratio of the focal length of the source and its aperture; therefore, it is a measure of the focusing angle of the source. B-mode ultrasound was used for targeting and sonication guidance. A linear ultrasound imaging probe (ATL L7–4 probe, 4 – 7 MHz, Philips, Bothell, WA, USA) was aligned with the HIFU focal plane, as shown in Figure 1(b – c) and was controlled by a Verasonics V1 system (Redmond, WA, USA).

Exposure parameters

Three sets of exposure parameters (Table 1) were used to generate volumetric lesions in n=21 hematoma samples. Two regimes utilized pulse durations typical for BH in soft tissues (10 and 2 ms) with 1% duty cycle and were applied in n=6 samples per parameter set (Khokhlova et al. 2011). The third sonication regime others have used for hybrid histotripsy utilizes sub-millisecond pulses (0.4 ms), which are noticeably shorter than BH pulses but much longer than typical CH pulses (Eranki et al. 2018), was applied to n=9 samples. The number of samples per BH regime was limited by the total available blood volume with CPD-SAGM collected in a regular donor blood container. Since the hybrid regime was less studied, the number of samples for this treatment was increased.

TABLE 1.

Exposure parameters for volumetric liquefaction of human hematoma phantom in BH and hybrid histotripsy regimes: F-number (F#), operational ultrasound frequency (f₀), pulse duration (t_pulse), equivalent number of pulses per point (N_p), duty cycle (DC), acoustic power in situ (W_hematoma), peak positive (P⁺) and negative pressure (P⁻) at the focus, focal shock amplitude (A_s) and time-to-boil (t_b).

	F#	f0, MHz	t pulse, ms	N p	DC	Whematoma, W	P+ / |P−| / As, MPa	tb, ms
BH	0.75	1.5	10	8	1 %	238.7	120 / 17 / 101	1.85
			2
Hybrid			0.4	11		424	148 / 21 / 151	0.55

For BH regimes, the necessary output power was calculated based on the required time-to-boil (t_b) in situ, which had to be less than the pulse duration (Khokhlova et al. 2018b). Specifically, using the combination of the heat transfer equation and weak shock theory, the required shock amplitudes in situ were calculated as follows (Canney et al. 2010; Hamilton and Blackstock 1998; Khokhlova et al. 2011):

$$A_s=\sqrt[3]{\frac{\Delta T{\text{c}}_{v}6{\rho }^2{c}_0^4}{\beta f_0{t}_b}},$$ #(1)

For the operating HIFU frequency f₀ = 1.5 MHz, the shock amplitude had to be higher than 57.5 MPa and 98.3 MPa for time-to-boil to be less than 10 and 2 ms, respectively. The temperature difference, ΔT, in Eq. (1) is the difference between the ambient temperature and 100 °C. Heat capacity per unit volume (c_v) in human blood clot was calculated as heat capacity per unit mass, previously measured by Nahirnyak et al., multiplied by the clot density ρ = 1060 kg/m³: c_v = 3.71 МJ/m³·K (Nahirnyak et al. 2006). Sound speed in human blood clot c₀ = 1585 m/s, its density ρ = 1060 kg/m³ and nonlinear parameter β = 4.15 were also taken from literature (Khokhlova et al. 2016a; Nahirnyak et al. 2006; Shung et al. 1984). In the above calculations, the thermal diffusion was considered to be negligible.

Note that the same output power was used for both BH regimes (Table 1), with the theoretical time-to-boil of 1.85 ms being within both 10-ms and 2-ms pulse lengths. This is because a shock amplitude lower than 100 MPa is achieved when the shock is not fully developed for the highly focused transducer used in this work. This regime is very sensitive to fluctuations in output acoustic power or inhomogeneities encountered in the acoustic path and is therefore not very reliable (Khokhlova et al. 2018b). The output power level required to achieve the necessary shock amplitude at the focus in the hematoma sample was determined based on previous acoustic characterization of this HIFU system via combined hydrophone measurements and modelling (Khokhlova et al. 2018b, Yuldashev et al. 2018, Khokhlova VA et al. 2018) in combination with nonlinear derating procedure (Bessonova et al. 2010; Canney et al. 2010):

$$W_{hematoma}=W_{water}{e}^{2\alpha D}$$ #(2)

W is the output acoustic power in water and in the hematoma producing the same focal waveform, α = 0.076 Np/cm is the attenuation coefficient of the clotted blood at 1.5 MHz (Shung et al. 1984) and D = 11 mm is the mean depth of HIFU focus in tissue. The equivalent source parameters for modelling based on wide-angle parabolic equation (WAPE) were as follows: 72.7 mm aperture, 60 mm focal length, 30 mm-diameter central opening, and a scale factor p₀/V₀ = 4.508 kPa/V for conversion from the pressure at the transducer surface to the applied voltage.

For the hybrid regime, the highest achievable power of the amplifier was used. Bubble activity initiation was verified experimentally by observing a bright hyperechoic region on the B-mode image after delivering a single BH pulse or several 0.4-ms pulses.

The numerically modelled in situ focal waveforms corresponding to the power levels used in each regime are shown in Figure 2. Parameters of these waveforms are provided in Table 1. For the hybrid regime (t_b = 0.55 ms), time-to-boil was calculated to be slightly longer than the pulse duration, which corresponded to the experimental observations of the bright spot on B-mode images.

FIGURE 2.

Simulated waveforms for BH and hybrid histotripsy regimes in the hematoma phantom, at the focus of the 1.5 MHz transducer. The waveforms were calculated using a wide-angle parabolic equation with an equivalent source method combining hydrophone measurements in water with nonlinear derating procedure.

Sonication geometry

Volumetric histotripsy lesions (3 – 4 mL in volume) were generated by continuous translation of the transducer focus along a square-wave trajectory (Fig. 3) in several transverse planes, starting distally and progressing proximally. The folding parameter d (i.e. the distance between the lesion lines) and spacing between the layers S were determined so that the nearest lines of the lesions effectively merged to produce homogeneous volumetric lesions. The choice was based on the preliminarily measured sizes and shapes of single lesions produced with a number of pulses Np (Table 1) from B-mode images and gross photographs (see the Results section) for each set of exposure parameters. For the BH regimes, the optimal number of pulses per point was selected as Np=8 based on the rapid increase of the void size with the number of pulses up to that number (Khokhlova et al. 2018a). The folding parameter d was chosen to be about 50% smaller than the diameter of the corresponding lesion “heads” to merge the lesions produced in adjacent lines. Focus translation speed, v, was calculated such that an equivalent number of pulses, Np, would be deposited per focus, assuming the same distance, d, between each single foci in a line as between the lines of lesions (Fig. 3): v=d/(t_pulse·100·Np). For the hybrid regime, because single lesions were much smaller, the optimal equivalent number of pulses per point, Np = 11, was determined by sonicating separate lines with varying focus translation speeds to produce continuous lesion lines at the fastest rate. The liquefaction rate, in cubic centimeters per minute, was estimated as the square of the line width multiplied by its length and focus translation speed.

FIGURE 3.

Focus translation trajectory for producing volumetric lesions.

Analysis of histotripsy outcome

Lesion size evaluation and gross analysis.

The length and width of the lesions (i.e. in z and x directions, see Fig. 3) were first measured based on B-mode images of the hypoechoic area that formed about 5 – 10 minutes following the exposure. This hypoechoic area corresponded to the liquefied part of the hematoma. After the treatment, samples were bisected first in the xy plane and then in the axial plane for gross observation and photography in all three dimensions. The width, length, and depth of the lesions (i.e. dimensions in x, y, and z directions, see Fig. 3) were also measured from these gross photographs and compared to those measured from B-mode images. The approximate lesion volume was then calculated as a product of the three measured dimensions.

Sample collection.

Following the treatment, 2 samples for each of the three treatment regimens were randomly selected for cytological and ultrastructural analysis. After the samples were bisected, the liquefied content was aspirated from the lesions using an 18G needle for subsequent light microscopy and SEM.

Small fragments of intact hematoma captured from its volume by an 18G fine needle immediately after recalcification and three hours later served as control untreated samples for both cytology and SEM.

Light microscopy.

One drop of the aspirated lysate was smeared on a glass slide, fixed with May-Grunwald solution, and stained using Wright-Giemsa stain. The smears were photographed at 40X magnification on a Zeiss Axio Imager light microscope (Germany) equipped with a color digital camera (AxioCam MRc5, 2584×1936 pixels resolution, Germany).

Scanning electron microscopy (SEM).

For ultrastructural analysis, a drop from each sample was placed on a 10 mm diameter round cover glass substrate, fixed using 2.5% glutaraldehyde solution (DC Panreac, USA) and dehydrated in ethanol of increasing concentration. Dehydrated samples were soaked in hexamethyldisilazane (Sigma-Aldrich, USA) and air-dried before mounting and gold sputtering (Buravkov et al. 2011). The choice of the glass substrate over carbon adhesive tape as a substrate was made based on preliminary observations of the ultrastructure of different substrates, so that they would not be confused for the fibrin structure. The lysate samples were examined and photographed using a JEOL JSM-6380LA Analytical Scanning Electron Microscope (Japan).

To examine individual debris within the lysate drop, as opposed to multiple fused layers of debris, the lysate was diluted either in phosphate-buffered saline (PBS) (PanEco, Russia) or in distilled water. The use of distilled water was expected to result in bursting of white blood cells and RBCs due to osmotic pressure. However, the structure of the fibrin network segments was expected to remain unaffected by dilution in water, and the cell sizes (6 – 8 μm in diameter) were significantly smaller than the largest residual fibrin fragment of interest. The samples obtained with dilution in PBS were also examined under SEM for comparison.

Prior to the histotripsy experiments, fresh human blood was diluted with PBS at concentrations of 1:25, 1:50, 1:75, 1:100, placed on a glass substrate, and prepared for subsequent SEM investigation. The results of these experiments provided the optimal dilution level resulting in complete separation of blood cells and no overlapping of them in SEM images, with spacing sufficient for separating larger fibrin network fragments (>70 μm) expected to be observed in liquefied lysate after the treatment.

After histotripsy exposures, a part of the liquefied content from each sample of the group was diluted with either PBS solution or distilled water at a concentration of 1:50 (the optimal dilution level defined as described above) and prepared for subsequent SEM analysis. Each SEM sample, approximately 8 – 10 mm in diameter, was scanned in its entirety and examined for fragments noticeably larger than RBCs (above 20 μm) to eliminate RBCs from sizing and focus on large fibrin net fragments only. These fragments were photographed (overall, 23 fields of view) and analyzed in ImageJ (NIH, Bethesda, MD). Each fragment was manually outlined and the “maximum Feret diameter” was used as a measurement of fragment size.

In order to define whether fine-needle aspiration influenced the ultrastructure of samples being observed on SEM, a preliminary experiment with human native blood was performed. A drop of human native blood was placed on a substrate and allowed to clot at 37°C. An 18G needle was used to aspirate part of the clotted drop. The ultrastructure of the blood clot collected by the needle was compared to the clot left on the substrate using SEM (Fig. 4). No apparent differences were observed. Both the structure of the fibrin network (indicated by white arrows in Fig. 4) and the shape of the cells (seen under low magnification, Fig. 4a,b) remained the same, which can be seen even under high magnification (x12 000). These results indicate that fine-needle aspiration did not alter the ultrastructure of the aspirated material.

FIGURE 4.

Illustration of the influence of fine needle aspiration on the ultrastructure of clot components. The content of the human blood clot: (a,c) aspirated by a needle; (b,d) left on a substrate. White arrows point to fibrin network. Scale bars: (a – b) 20 μm, (c – d) 1 μm.

RESULTS

Single lesions

B-mode images of single lesions produced by delivering N_p = 8 pulses of the BH regimes and lines of lesions obtained with N_p = 11 pulses per sonication point of the hybrid histotripsy regime are shown in Figure 5. The single lesions induced by BH have the typical “tadpole”-shape (Fig. 5,a – b). Delivery of shorter pulses resulted in axially shorter lesions (both heads and tails were shorter), with the narrower “head” being almost the same width as the “tail” (Fig. 5,c). Sonication geometry parameters and focus translation speed (Table 2) were determined by the size of these lesions (Fig. 5) to achieve effective merging of BH single lesions (primarily the “heads”) and lesion lines for the hybrid regime.

FIGURE 5.

B-mode images of single lesions (HIFU incident from the left) induced by (a) eight 10-ms pulses, (b) eight 2-ms pulses and (c) eleven 0.4-ms pulses. Scale bar is 5 mm.

TABLE 2.

Parameters of experimentally performed large-volume hematoma treatments in BH regimes and suggested hybrid histotripsy.

	t pulse	Focus translation speed υ	Max lesion volume	Treatment time	Max liquefaction rate
BH	10 ms	0.45 mm/s	3.34 cm3	11 min	0.3 mL/min
	2 ms	1.1 mm/s	3.56 cm3	5.23 min	0.68 mL/min
Hybrid	0.4 ms	5 mm/s	3.67 cm3	1.4 min	2.62 mL/min

In-treatment ultrasound imaging and gross observations

Untreated phantoms (Fig. 6a) appeared mildly echogenic on B-mode images, in agreement with literature on the sonographic appearance of hematomas (Wicks et al. 1978). The echogenicity stems from scattering by red blood cells trapped within the fibrin network. During the exposure (10 ms regime), the treated area appeared as a bright hyperechoic region due to the presence of vapor and gas bubbles (Fig. 6b). As the bubbles dissolved after the treatment, the liquefied lesion gradually turned hypoechoic over 10 minutes (Fig. 6c).

FIGURE 6.

B-mode imaging of a volumetric BH lesion produced with 10 ms pulses (HIFU incident from the left): (a) before the treatment, (b) during the treatment of the distal layer with focus (F) moving transversely to the transducer axis, and (c) 5 minutes after treatment. Black dashed line in (b) indicates the trajectory of the transducer focus movement. Scale bar is 10 mm.

B-mode ultrasound images of volumetric lesions and gross pictures of the voids after sample bisection for each of the utilized regimes are shown in Figure 7. Note that the B-mode images (Fig. 7, a – c) are shown in the axial plane of the HIFU transducer (xz plane containing the transducer acoustic axis, Fig. 2) and gross photographs (Fig. 7,d – f) in the transverse plane (xy plane, perpendicular to the transducer acoustic axis, Fig.2). 5 – 10 minutes after the exposures (Fig. 7, a – b), the volumetric lesions appeared as hypoechoic areas of completely liquefied material. The white dashed rectangles in Figure 7 outline the lesions, as defined by the B-mode image taken after the bubbles dissolved. Note that the image shown in Figure 7c was taken before the treated area turned fully hypoechoic to illustrate that the bubbles were pushed towards the lesion “tails” and dissolve later than those in the proximal part of the liquefied volume. The irregular “tadpole” of individual single BH-lesions resulted in a volumetric homogenized proximal area and distal hypoechoic regions of discrete “tails” (marked by white arrows, Fig. 7a,b). The volume disrupted by these discrete “tails” was not included in the overall volume estimation.

FIGURE 7.

(a – c) B-mode images (HIFU incident from the left, along the z axis) and (d – f) bisection photos (HIFU incident perpendicularly to the image plane) of the volumetric lesions induced by BH with 10-ms pulses (left column) and 2-ms pulses (middle column) and by the hybrid regime with 0.4-ms pulses (right column). White dashed boxes outline the volumetric lesions. In the B-mode images (a – c), narrow dotted lines indicate the layers of transducer focus translation. In the bisection photos (d – f) narrow dotted lines indicate transducer focus translation trajectory in each layer. Scale bar is 5 mm.

The samples were bisected along the proximal edge of the lesion in xy plane (Fig. 7d – f). The lesion volume measurements made from B-mode and gross images are provided in Table 2. The size of the lesions along the x axis agreed well with those obtained from the corresponding B-mode images. No apparent differences in lesion contents between different pulsing protocols could be discerned by gross observation.

Liquefaction rate

The liquefaction rate for each of the three sonication regimes was calculated from the measured liquefied void size and sonication time (Table 2). In this study, the highest thrombolysis rate of 2.62 mL/min was achieved with sub-millisecond pulses (0.4-ms).

Cytological analysis

Regardless of the time point after blood recalcification, the control untreated hematoma samples stained with Wright-Giemsa stain appeared as densely packed RBCs and leukocytes under light microscopy (Fig. 8a). Regardless of the histotripsy exposure parameters, smears of the lesion lysate revealed nonviable shadows of erythrocytes and destroyed white blood cells with condensed nuclei and no cell membrane (Fig. 8,b – d). The pink-tinged background seen after histotripsy exposure indicates hemolysis caused by the treatment. No influence of treatment protocol was observed on the lysate contents under light microscopy.

FIGURE 8.

Light microscopy images of (a) intact hematoma and (b – d) hematoma lysate treated with BH (boiling histotripsy) pulses: (b) 10 ms, (c) 2 ms; and with hybrid regime using 0.4-ms pulses (d). Scale bar: 20 μm. Black arrows point at leukocytes: (a) intact, (b – d) destroyed. Treatment-induced hemolysis appears as tinted background (b – d).

Scanning electron microscopy of untreated hematoma and histotripsy lysate

Two control samples of untreated hematoma were aspirated for SEM: one was taken immediately after recalcification was visually complete (Fig. 9a,c), and the other three hours later (Fig. 9b,d). As seen in Figure 9, the ultrastructure of visually clotted blood had changed over the three hours after recalcification; fibrin fibers between RBCs were more clearly seen at the latter time point. These changes were in agreement with the literature (Bernal et al. 2012) as the elastiс properties of the large volume clot were shown to be changing over the first 3 hours after the start of coagulation.

FIGURE 9.

SEM (scanning electron microscopy) control images of intact hematoma immediately after recalcification (a,c) and three hours after recalcification (b,d). Scale bars: (a – b) 5 μm, (c – d) 1 μm. White arrows point at fibrin fibers.

In the treated hematoma samples (Fig. 9), only multilayered fibrin network debris was visible, regardless of the treatment protocol (Fig. 10). Higher magnification showed layered fragments of destroyed fibrin network covered with cell remnants (Fig. 10,d – f).

FIGURE 10.

SEM (scanning electron microscopy) images of hematoma lysate treated with BH (boiling histotripsy) pulses: (a,d) 10 ms, (b,e) 2 ms; and with hybrid regime using 0.4-ms pulses (c,f). Scale bars: (a – c) 10 μm, (d – f) 2 μm.

Scanning electron microscopy of diluted lysate

To separate the residual fibrin network debris while maintaining the isotonicity of the blood components, a group of lysate samples was diluted with PBS. As a result, individual residual clot components were more clearly observed. However, occasional salt crystals were detected leading to adhesion of initially separated clots to those crystals, thus hindering the analysis of the debris (Fig. 11a,d).

FIGURE 11.

SEM (scanning electron microscopy) images of: (a,d) salt crystal artefacts with aggregated clot debris after dilution of the lysate in PBS; RBCs (b,e) and ultrastructure of fibrin fibers (c,f) after dilution in PBS (b – c) or in distilled water (e – f). Scale bars: (a) 2 μm, (b – c, e – f) 1 μm, (d) 5 μm.

To avoid the aggregation of clot segments with salt crystals, the lysate was diluted with distilled water. As expected, the use of distilled water led to bursting of blood cells (Fig. 11b,e), whereas fibrin ultrastructure present in the lysate remained the same as that observed in samples diluted with PBS, even under magnification of x13000 (Fig. 11c,f).

Dilution of the lysate with distilled water allowed for the evaluation of the size of individual fibrin network segments (Fig. 12a). Table 3 shows size distribution of the residual fragments larger than 20 micrometers. Regardless of the histotripsy treatment protocol, the largest of the fibrin segments that remained after the exposures were 150 micrometers in length and are shown in Figure 12(b,c).

FIGURE 12.

(a) Gross SEM (scanning electron microscopy) image of the lysate diluted with distilled water at a concentration 1:50; (b – c) SEM images of the largest fibrin network fragments present in the lysate diluted with distilled water after the treatment. Scale bars: (a) 50 μm, (b – c) 20 μm.

TABLE 3.

Large (>20 μm) fragments size distribution in the diluted lysate of the test group samples.

Sizes	20 – 50 μm	50 – 100 μm	100 – 150 μm
Number of fragments	17	7	4

DISCUSSION

The ultimate goal of this study was to optimize histotripsy regimes for liquefaction of large volume hematomas in terms of maximizing the thrombolysis rate while providing means for unobstructed fine needle aspiration, as evaluated from the residual fragment sizes. We have evaluated the performance of different histotripsy treatment protocols in the liquefaction of freshly clotted large volume human blood serving as an in vitro hematoma phantom. Specifically, two of the typical BH regimes involving longer (10 ms) and shorter (2 ms) pulses and a hybrid histotripsy regime using 0.4 ms pulses were used (Eranki et al. 2018) in combination with continuous HIFU focus translation along a square-wave trajectory. The optimal spacing between the meanders and layers was defined based on the geometric parameters of single lesions (for BH regimes) and the width of lesion lines (for a hybrid regime) induced by each pulsing protocol.

All three regimes successfully generated fluid-filled voids in gelatinous hematoma phantoms within clinically relevant times (3 – 4 mL within 2 – 11 mins). The hybrid histotripsy regime achieved the most uniformly shaped void, based on the absence of discrete clot filaments or lesion “tails” seen in B-mode images of the lesions, and the highest liquefaction rate of 2.62 mL/min. This thrombolysis rate was comparable to the clinical rates of thermal HIFU ablation of soft tissue (Ikink et al 2015; Illing et al. 2005) and was more than five times higher than that achieved previously in bovine blood clots for combined CH and BH raster treatment at 1.5 MHz frequency (Khokhlova et al. 2016). In the same study, a slightly higher thrombolysis rate of 2.8 mL/min was achieved using a 1 MHz transducer, which required higher acoustic power. The inverse dependence of the liquefied lesion size on frequency is well documented for both BH (Khokhlova et al. 2011) and other histotripsy methods, e.g. microtripsy (Lin et al. 2014a). For example, a thrombolysis rate of 16.6 mL/min (Gerhardson, et al. 2017) was achieved with transcranial microtripsy at a frequency of 250 kHz. However, to facilitate the efficient low frequency microtripsy approach, a highly focused hemispherical HIFU transducer was required in that study which is not feasible for the majority of intra-abdominal and retroperitoneal hematomas.

Continuous translation of the transducer focus within targeted region was performed in this study by mechanical manipulation of the sample with a 3-D positioning system. Therefore some time was spent on switching the focus position between the adjacent meanders and layers. These stops were not included in the treatment time calculation as they could be avoided by the use of electronic steering of the focus or the use of a more advanced positioning system (e.g. a robotic arm).

In an additional set of studies, the ability of a less focused 1.5 MHz source (F-number = 1) to liquefy hematoma phantoms with the hybrid histotripsy regime was also examined but proved impossible, likely due to insufficient shock amplitude and thus heating rate at the focus, even at the highest power achievable. Recent experimental studies using human blood clot phantoms (Khokhlova et al. 2016b; Khokhlova et al. 2018b; Rosnitskiy et al.) and estimations based on the Eqs. (1) – (2) show that initiation of boiling within one millisecond at a working frequency of 1.5 MHz requires in situ shock amplitude higher than 130 MPa and, therefore, cannot be produced by transducers with the F-number of 1 and over as the corresponding focal shock amplitudes saturate at a lower value. A higher HIFU frequency could reduce the shock front amplitude required for the hybrid histotripsy regime, e.g. a HIFU transducer with an F-number of 1 would enable initiation of boiling within one millisecond at frequencies of 4 MHz and higher, A_s being less than 90 MPa. However, higher frequencies would result in smaller single lesion volume and reduce the thrombolysis rate (Khokhlova et al. 2011).

One limitation of this study is the difference between the envisioned clinical implementation of the method and its laboratory implementation. First, in a clinical setting the HIFU beam will be attenuated and aberrated by the overlying inhomogeneous tissues and potentially by the inhomogeneities in the structure of the hematoma itself. The aberration will distort the HIFU beam, shift the focus, and may prevent shock formation at the intended point. These effects have to be mitigated using the emerging methods of aberration correction based on pre-operational CT scans or backscattering of nonlinear pulses (Martin et al. 2020; Suomi et al. 2018; Thomas et al. 2021). The overlying tissues may also influence the quality of B-mode imaging guiding the liquefaction process, as shown previously for BH ablation of other abdominal targets (Khokhlova et al. 2019). Further, in a clinical setting, the spatial translation of the HIFU focus would have to be realized either by mechanical translation of the transducer or by electronic steering of the focus.

Following the treatment, fine-needle aspirations of the lysates were taken from 6 out of 21 samples (2 samples per regime) for subsequent cytological analysis and scanning electron microscopy. Complete destruction of blood cells was observed in all samples, and there was no apparent dependence of debris structure on pulsing protocol. Preliminary studies demonstrated that fine-needle aspiration has no visually detectable effect on the clot ultrastructure, suggesting that it adequately represented the contents of the obtained lesion.

SEM images of the control untreated samples revealed clot ultrastructure similar to those of control samples presented in literature (Bester et al. 2018; Janis et al. 2002; Sutton et al. 2013). The liquefied lysate appeared as multiple fused layers of destroyed fibrin fragments covered with cell remnants making it impossible to define the sizes of individual lysate fragments. Optimal separation of the debris prior to SEM analysis was achieved by diluting the lesion content with distilled water to a final concentration of 1:50. This simple method allowed for manual sizing of the larger residual fragments, although it led to bursting of the blood cells. The choice of distilled water over isotonic PBS was made in an effort to avoid aggregation of the debris to salt crystals. The fibrin network debris was shown to preserve the same ultrastructure in distilled water as in PBS. The predominant part of the fragments was smaller than 20 micrometers which corresponded to the results presented in the literature (Khokhlova et al. 2016a; Li et al. 2020). Sizing of the larger debris (>20 micrometers) showed that the largest fibrin fragments observed were 150 micrometers in length which would most definitely not hinder a fine-needle aspiration in clinical conditions via commonly used needles of up to 23G (opening of 160 μm and larger).

These results show the potential of using histotripsy with continuous focus translation to rapidly and non-invasively break down large hematomas within clinically relevant time into the debris sufficiently small for subsequent fine-needle aspiration. The hybrid sub-millisecond histotripsy method resulted in a faster liquefaction rate compared to BH regimes.

CONCLUSIONS

This work aimed to investigate the liquefaction of large human hematoma by continuous delivery of the histotripsy pulses of varying length with continuous HIFU focus translation and to refine SEM methods to evaluate the exposure outcomes. Three types of histotripsy approaches (BH with 10 ms and 2 ms pulses, and a hybrid sub-millisecond regime) were used for treating large human hematomas in vitro. All the regimes yielded volumetric lesions within clinically relevant time, with the highest thrombolysis rate and the most spatially-uniform void achieved by the hybrid pulsing protocol. SEM and cytological analysis of the lysate confirmed complete cells disintegration for all the regimes. Dilution of the lysate in distilled water allowed for manual sizing of the individual debris, most of them being smaller than 20 μm, and the largest was 150 μm in length, thus unlikely to hinder fine-needle aspiration.