Design, characterization and evaluation of a laser-guided focused ultrasound system for preclinical investigations

Background The clinical applications of transcranial focused ultrasound continue to expand and include ablation as well as drug delivery applications in the brain, where treatments are typically guided by MRI. Although MRI-guided focused ultrasound systems are also preferred for many preclinical investigations, they are expensive to purchase and operate, and require the presence of a nearby imaging center. For many basic mechanistic studies, however, MRI is not required. The purpose of this study was to design, construct, characterize and evaluate a portable, custom, laser-guided focused ultrasound system for noninvasive, transcranial treatments in small rodents. Methods The system comprised an off-the-shelf focused ultrasound transducer and amplifier, with a custom cone fabricated for direct coupling of the transducer to the head region. A laser-guidance apparatus was constructed with a 3D stage for accurate positioning to 1 mm. Pressure field simulations were performed to demonstrate the effects of the coupling cone and the sealing membrane, as well as for determining the location of the focus and acoustic transmission across rat skulls over a range of sizes. Hydrophone measurements and exposures in hydrogels were used to assess the accuracy of the simulations. In vivo treatments were performed in rodents for opening the blood–brain barrier and to assess the performance and accuracy of the system. The effects of varying the acoustic pressure, microbubble dose and animal size were evaluated in terms of efficacy and safety of the treatments. Results The simulation results were validated by the hydrophone measurements and exposures in the hydrogels. The in vivo treatments demonstrated the ability of the system to open the blood–brain barrier. A higher acoustic pressure was required in larger-sized animals, as predicted by the simulations and transmission measurements. In a particular sized animal, the degree of blood–brain barrier opening, and the safety of the treatments were directly associated with the microbubble dose. Conclusion The focused ultrasound system that was developed was found to be a cost-effective alternative to MRI-guided systems as an investigational device that is capable of accurately providing noninvasive, transcranial treatments in rodents.


Background
The use of therapeutics specifically for the treatment of brain disorders is impeded by the presence of the blood-brain barrier (BBB). The BBB's primary role is to regulate the exchange of substances between the blood vessels and the underlying tissues, protecting the brain from the entry of potentially harmful compounds [1]. Because the BBB's protective mechanisms are highly effective, it is a major obstacle in the effort to deliver drugs and other agents systemically to the brain. This has led, in turn, to the development of various techniques including direct intracranial administration of agents [2] or the use of systemically administered agents that can transiently alter the BBB's permeability [3]. These techniques, however, are either invasive, with associated risks for the patients, or they suffer from poor spatial and temporal control [1].
Focused ultrasound (FUS) is an emerging technology for the treatment of brain disorders [4]. Earlier investigated applications, currently in clinical use, include thermal ablation of uterine fibroids [5], bone metastases [6] and breast cancer [7]. Most recently, FUS was approved by the FDA for the treatment of essential tremor (ET), a debilitating neurological condition involving dysfunctional neural circuits that is normally treated with surgically-implanted deep brain electrodes. The noninvasive MRI-guided FUS (MRgFUS) procedure involves transcranial exposures for the ablation of the ventral intermediate (VIM) nucleus, located within the thalamus [8]. The ability to safely target the VIM nucleus for ET has created the opportunity to apply FUS to other brain-related applications, including enhanced therapeutic delivery to the brain. To date, a number of preclinical studies have reported that MRgFUS can safely and transiently open the BBB, enabling a wide range of therapeutic agents to enter the brain at a targeted location [9,10], including adult stem cells [11]. Whereas MRgFUS can be very effective for the stated purpose, these preclinical systems are often expensive to purchase and operate. Treatments can also be prohibitively long and are potentially restricted to imaging centers within the research institution where studies are being carried out. While in some cases MRI-guidance is necessary for accurately targeting discrete locations in the brain, often, for general mechanistic studies, it is not required.
In this paper, we present a custom-FUS system for noninvasive, transcranial treatments in small animals. The system is portable, employing laser guidance with a 3D positioning stage for precise targeting of the focused beam without MRI. We designed a custom cone that is filled with degassed water for direct coupling of the transducer to the head of the animal. The degassing setup was also custom-built for the system. Computer simulations were performed to characterize the pressure field generated by the transducer with and without the cone, characterize the location of the focus, and predict transmission losses across the skull over a range of animal sizes. The approximate position and dimensions of the focus were then validated using high-powered, continuous FUS exposures in hydrogel phantoms for the generation of visibly opaque thermal lesions. Exposures were also performed in ex vivo tissue to validate the targeting of the custom, laser guidance apparatus. Finally, proof-of-concept studies were carried out in vivo for opening the BBB in healthy Sprague-Dawley rats. The experiments were designed to confirm the safety and efficacy of the system, as well as its applicability for mechanistic investigations of the process of BBB opening.

System design
The major component of the system (see Fig. 1) is a spherical, single-element FUS transducer (Sonic Concepts, Bothell, WA, USA) operating at a center frequency of 500 kHz and a frequency bandwidth ranging from 400 kHz to 600 kHz. The curvature radius and aperture diameter are 63.2 mm and 82 mm, respectively, yielding an F-value of 0.77. The transducer possesses a focal width (− 6 dB) of 2.36 mm and focal length (− 6 dB) of 13.50 mm. A clear plastic cone was designed and constructed to facilitate coupling of the transducer to the target. The height of the cone is 26.6 mm from the exit plane of the transducer housing to the exit plane of the cone. An acoustically transparent (0.05″) silicone membrane (McMaster-Carr, Elmhurst, IL) seals the exit plane of the cone, allowing for direct coupling between the transducer and the target. The membrane can be remotely inflated or deflated to offset the position of the fixed focal distance within the designated treatment region. The cone is filled with water that is degassed using an in-line degassing membrane (PermSelect, Ann Arbor, MI, USA) powered by an external vacuum pump. The transducer is driven by an acoustic amplifier (TPO-102, Sonic Concepts, Bothell WA, USA) via a standard impedance matching system (50 Ω). The amplifier supplies a maximum output power of 150 W. The graphic user interface (GUI) allows for varying the power (at increments of 0.1 W), the pulse width from 10 μs to 1 s (at increments of 1 μs), the pulse repetition frequency from 1 kHz to 0.1 Hz (at increments of 1 kHz), and the treatment duration at increments of 0.1 s. Varying the levels of the different exposure parameters is carried out by selecting the appropriate parameter on the touch screen of the GUI and using a digital dial to increase or decrease its value at their corresponding increments. A custom positioning system (Velmex, Bloomfield, NY, USA) was assembled using three aluminum alloy axes to allow for independent movement in the x (side-to-side), y (forward-and-back) and z (up-and-down) directions relative to the transducer. Additional custom-designed components manufactured specifically for the system by the company include a 30.5 cm aluminum alloy bracket arm (attached to the vertical z-axis) . Each axis has an independent travel length of 12.7 cm at a resolution of 1 mm. A custom 3D-printed adapter (Makexyz, Austin, TX, USA) was designed to connect the transducer to a ball-joint camera mount (Sinvitron, Hong Kong, China), with the latter attaching to the bracket arm. The combined axes and the ball-joint mount allow for a total of five degrees of freedom in movement, rendering the positioning system sufficiently flexible for a variety of treatment applications. A laser-based apparatus was designed for targeting the exposures in the (x, y)-plane. The apparatus was custom-made with a 3D printer (LulzBot TAZ 5, Aleph Objects, CO) for a single off-the-shelf laser (Fig. 2a, b). The 650 nm dot laser (DIYMall, Shenzhen, China) was operated at 5 V. The apparatus was designed such that it can be placed onto the exit plane of the cone using ultrasound coupling gel. Once targeting is determined during treatment planning, the apparatus is then removed. Demonstration of the suitability and accuracy of this procedure was carried out using thermal ablations in ex vivo tissue samples (Fig. 2c, d) [12]. The laser apparatus was used for the in vivo treatments (below), employing stereotactic coordinates for targeting specific regions within the brain.

Transducer calibration
A radiation force balance-based power meter (Ohmic, Easton, MD) was used for calibrating the FUS transducer. Continuous exposures were carried out at 5 W, 7 W, 12 W, 15 W, 17 W and 20 W (n = 5 for each power) as indicated on the GUI of the amplifier. Measurements were carried out in degassed water to minimize the effects of acoustic cavitation. Exposures were maintained until the output power stabilized. An average efficiency (measured power/applied power × 100) of 86% was observed over the range of powers that were evaluated. All powers cited for the remainder of the study were those calibrated at this level of efficiency.

Simulations of the acoustic pressure fields
Computer simulations were carried out to characterize the acoustic pressure fields generated by the FUS transducer. These were performed under four conditions: in free-field for the transducer alone; in free-field for the transducer with the cone but without the membrane; and for the transducer with the cone, including the acoustically transparent membrane, both inflated and non-inflated. The geometry of the bolus was simulated as an arc with a height corresponding to the height of the inflated membrane in the experimental setup for the measurements carried out to validate the simulations (see below). These four conditions were considered for two reasons: (i) to determine if the presence of the cone affected the ultrasound field, and (ii) to demonstrate the importance and advantage of using degassed water encased within the cone that was sealed by the membrane. The simulations were performed using a two-dimensional axisymmetric model developed in COMSOL version 5.3 (COMSOL, Burlington, MA). COMSOL solves the nonlinear wave propagation equation to calculate the magnitudes of the acoustic pressure fields. With the commonly used assumption that the pressure generated by the transducer can be represented as a simple harmonic sinusoidal wave, the Helmholtz equation [13] can be reduced as: where p is the pressure (Pa), ρ is the density of the medium (kg/m 3 ), ω is the angular frequency (rad/s) and c is the speed of sound (m/s). The simulations were carried out at the center frequency of the transducer at 500 kHz in continuous mode. The material properties of the different components of the system were obtained from the COMSOL materials library and additional sources [14,15] and appear in Table 1. These include the cone, made of polyvinyl chloride (PVC) and the acoustically transparent membrane, made of silicone rubber.

Localization of the focus
In order to verify the position of the focus of the FUS transducer, exposures were carried out in ultrasound hydrogel phantoms. The phantoms were prepared as previously described [12]. Briefly, bovine serum albumin (BSA) was dissolved in a polyacrylamide hydrogel solution at a percentage of 7% (w/w) and allowed to set. Phantom samples (30 mm in height, width and length) were treated at their approximate geometrical center, based on our simulation results. The exit plane of the cone was positioned just above the surface of the phantoms, with the bolus deflated. Degassed ultrasound coupling gel was added between the bolus and phantom. The bolus was then inflated remotely with a syringe (filled with degassed water) so that the bolus was in direct contact with the phantom with a thin layer of gel between them. Ultrasound exposures were carried out at 80 W in continuous mode for a duration of 20 s (n = 6) to create a thermal lesion. Scaled images of the lesions within the phantoms were captured with a digital camera. Lighting provided behind the phantoms was used to delineate the lesions and determine their location relative to the exit plane of the cone.

Hydrophone measurements of the acoustic pressure fields
In order to validate the simulations, we carried out hydrophone measurements using a custom setup. Measurements were carried out in a water tank (51 cm × 27 cm × 32 cm) filled with deionized water. A needle hydrophone (Onda, Sunnyvale, CA, USA) was positioned in the approximate region of the focus, across from the FUS transducer. The hydrophone was held in position using a custom holder (Solidworks, Waltham, MA, USA) and fabricated using a 3D printer (LulzBot TAZ 5, Aleph Objects, CO).Determining the exact position of the focus required an iterative process, where the transducer was moved in all three dimensions (x, y and z) to maximize the measured voltage. Measurements were recorded using a digital oscilloscope (Tektronix, Beaverton, OR, USA) at 0.5 W, 0.7 W, 1.0 W, 1.2 W and 1.5 W. The ultrasound exposures were carried out in continuous mode (n = 5 for each power). As in the simulations, measurements were recorded in free field for the transducer alone, the transducer with the cone, and the transducer with the cone and acoustically transparent membrane both inflated and non-inflated.

Simulating transmission across the skulls
In order to investigate the effect of the skull on our treatments in rats, we first carried out simulations using our COMSOL model (described above). In our model, we positioned a rectangular section of skull in the immediate pre-focal region, which was representative of in vivo conditions. The simulation model incorporated the speed of sound, density and attenuation coefficient for skull bone [16] ( Table 2). The peak acoustic pressure at the focus was determined with and without the presence of the skull at the center frequency of the transducer (500 kHz). Values of the acoustic pressures were simulated over a weight range from 80 to 675 g, based on the weight range of the rats from whom skulls were scanned with computed tomography (CT) for thickness measurements. The pressures were normalized to values without skulls.

Measuring transmission across the skulls
The hydrophone setup (described above) was used to determine the acoustic transmission across the skulls of the rats. Skulls were harvested from a range of different sized female, Sprague-Dawley rats (67.4 g to 630 g; n = 3 per weight). Rats were euthanized, and their skulls removed and cleared from residual brain tissue with a mild enzymatic detergent. The skulls were then rinsed and degassed in a custom vacuum chamber prior to measurements [17]. A custom holder was designed for the skulls (Solidworks, Waltham, MA, USA) and fabricated using a 3D printer (LulzBot TAZ 5, Aleph Objects, CO). Measurements were carried out in the same water tank as described above (Fig. 3a).
For the baseline reference measurements without the presence of a skull, the acoustic attenuation was measured for the water path, between the transducer and the hydrophone. Measurements were recorded on a digital oscilloscope (described above). At 1 and 2 Watts, consistent measurements were observed, however, below the threshold of acoustic cavitation in the beam path. The FUS exposures were carried out in continuous mode (n = 5 at each power). For measurements with the skulls, they were positioned such that the parietal bone (Fig. 3b) was located just in front of the hydrophone tip (as would occur during an in vivo treatment), half way between the bregma and the lambda, and half way between the sagittal suture and the outer edge of the skull.

In vivo treatments for opening the BBB
Animal experiments were carried out according to an approved protocol by the University of Maryland School of Medicine Institutional Animal Care and Use Committee (IACUC). For the treatments, female Sprague-Dawley rats were used over a range of weights (see below). All procedures were carried out under anesthesia using inhalation isoflurane (~ 2.5%, O 2 L/h). The animals were shaved using electric clippers, followed by a brief application of a depilatory cream to remove all residual hair for maximal coupling. The animals were then positioned in a stereotactic frame. The FUS transducer was positioned over the head of the animal using the laser guidance apparatus. Once the location of the initial treatment location was set, the laser apparatus was removed, and the transducer coupled to the animal's head using ultrasound gel (Fig. 4a).
With the transducer in position, animals were injected via a tail-vein catheter with a dose of commercially obtained monodisperse (4-5 μm) microbubbles (MBs) (Advanced Microbubbles Laboratories, Boulder, CO, USA). MBs were diluted with normal saline to a final volume of 400 μL prior to infusion. Immediately following a single injection of MBs, the FUS treatments were started. For the first two sets of animals, six individual locations were treated in a 2 × 3 raster pattern (Fig. 4b), covering the entire head with a lateral spacing of 2.5 mm, the approximate radial diameter (− 6 dB) of the focus. Exposures were carried out for 2 min per location using a 10% duty cycle and pulse repetition frequency of 10 Hz ( Once the treatments were finished, the group of smaller animals (113 g to 122 g) was injected with a single dose of 4 mL/kg of Evans blue dye (EBD) via the tail-vein at a concentration of 4% in 0.9% phosphate buffered saline (PBS, pH 7.4) [18]. Following a 30 min period to enable uniform circulation of the EBD [18], the animals were perfused with 30 mL PBS; (pH 7.4) followed by 30 mL of a 4% paraformaldehyde (PFH) solution in PBS. The animals were then euthanized, and their brains removed and sectioned at 1 mm slices using a brain matrix slicer (Zivic Instruments, Pittsburgh, PA, USA). Observations were carried out for the presence of EBD, as a positive indicator of BBB opening. Digital images were captured of each slice on both sides.
FUS exposures for BBB opening were carried out in a second group of larger animals (234 g to 250 g). BBB opening was validated by MRI as previously described [11]. Instead of EBD, a gadolinium-based contrast agent was injected systemically (0.8 mL/kg; Omniscan, GE Healthcare, Princeton, NJ). The animals were then scanned by T1-weighted (T1w) images for the presence of the contrast agent, which appears as a hyperintense signal.
All in vivo MRI scans were performed using a Bruker BioSpec 70/30USR Avance III 7T horizontal bore MR scanner (Bruker, Billerica, MA, USA) equipped with a BGA12S  A third set of animals (207 g to 245 g) was treated using three different MB doses (100 µL, 50 µL and 5 µL). Evaluation of BBB opening was carried out by T1w contrastenhanced MR imaging (described above). Two individual locations were treated at a spacing of 4 mm along the anterior-posterior axis in the left hemisphere at 0.3 MPa for 2 min each. The contralateral right hemisphere did not receive treatment and was used as an internal control. Following treatments, animals were scanned by MRI, employing three separate scanning modes: T2w (for observations of edema), T2* (for observations of microhemorrhage) and T1w (for observations of BBB opening, following the administration of contrast).
Immediately following the MRI scans, animals in the last group were perfused and euthanized, and their brains removed. These were fixed, sectioned, mounted on glass slides and stained with hematoxylin and eosin (H&E) for histological analysis. Observations were carried out with a light microscope at 10× and 20× magnification and representative images were captured using a digital slide scanner (Keyence, Itasca, IL, USA). Observations of the section were carried out for microhemorrhage in the form of red blood cell extravasation, as well as other manifestations of structural damage as previously described by us [11,19].

Localization of the focus
The FUS exposures in the hydrogel phantoms were found to generate sufficiently high heat to denature the albumin, which becomes optically opaque and visible to the naked eye. Using the scaled images, the mean distance from the center of the lesions to the exit plane of the cone was found to be 13.97 ± 0.29 mm (n = 6). A representative image of the setup and a thermal lesion appear in Fig. 5.

Simulations of the acoustic pressure fields
The position of the focus was determined according to the location of the peak pressure. This was found to occur at 14.1 mm from the exit plane of the cone and was the same for all four conditions evaluated (Fig. 6). The highest of these pressures was with the cone and inflated membrane. The results for the other three conditions were normalized to this value. The acoustic pressure at the focus for free field conditions (Fig. 6a) was found to be 19% of that with the inflated membrane (Fig. 6d). With the cone only, the peak pressure was 18% (Fig. 6b). Simulations with the cone and non-inflated membrane showed the acoustic pressure at the focus to be 70% of that with the cone and membrane inflated (Fig. 6c).

Hydrophone measurements of the acoustic pressure fields
As for the simulations, measurements with the inflated membrane were used to normalize the other three sets of measurements at each applied power. The measurements were generally found to be in accordance with the simulations when using the cone and membrane, both inflated and uninflated. Without the cone and/or membrane, disparities between the simulations and measurements were relatively greater. The results of the simulations and measurements appear in Fig. 6e.

Transmission simulations across the skulls
We employed our simulations to determine the relative changes in transmission of ultrasound energy across the rat skull that will occur over the range of animal sizes that are typically used in our studies. In preliminary experiments, we determined the relationship between the size (e.g., weight) of a rat and the thickness of its skull using a small animal CT scanner in order to measure the skull thickness (data not shown). The simulations showed the transmission to be proportionally lower as the size of the rats increased (Fig. 7).

Transmission measurements across the skulls
Hydrophone measurements were carried out with and without skulls positioned in the pre-focal region at 1 W and 2 W. Similar to the results of the simulations, an inverse relationship was found between the weight of the animals from which the skulls were harvested and the measured transmission. The transmission values were normalized to the measurements taken without a skull. The results, combined with the simulations, appear in Fig. 7.

In vivo treatments
In the first two sets of animals, a range of acoustic pressures was evaluated in order to determine the acoustic pressure threshold for opening the BBB for each animal size. Validation of BBB opening was determined by post-treatment visualization of either EBD or gadolinium contrast in the brain [11]. In the group of smaller animals, the acoustic pressure threshold for BBB opening was found to be 0.2 MPa (Fig. 8a). In the group of larger animals, the acoustic pressure threshold was 0.3 MPa (Fig. 8b).
In the third set of animals, BBB opening was assessed using T1w MR imaging at each of the three MB doses evaluated (100 µL, 50 µL and 5 µL). BBB opening was greatest at the MB dose of 100 µL, where a large, hyperintense region was observed that covered almost the entire left hemisphere. Microhemorrhage was evident as seen in hypointense signals in the T2* MR images. Less pronounced T1 enhancement was observed at the MB dose of 50 µL. T2* hypointensity was also less evident, indicating a reduction in microvasculature damage. At a MB concentration of 5 µL, T1 hyperintensity was further reduced. However, no T2* hypointensity was observed ( Fig. 9). At the 5 µL dose, histological sections showed no evidence of structural damage to the treated brains. Microhemorrhage was observed at 50 µL and found to be more pronounced at 100 µL (Fig. 10).

Discussion
In this paper, we describe the development of an effective FUS system for noninvasive transcranial treatments in small animal models. Typically, these treatments are carried out using expensive and comparatively hard to access MR-guided systems [20]. However, for many of the investigations that will be required to develop and optimize these treatments in order to expand their application into the clinical realm, we have shown that this can be done without MRI. At relatively little cost, the system was put together using an off-the-shelf transducer and amplifier with other parts, either purchased or produced using an in-house 3D printer. This included a simple laser guidance apparatus that we validated and demonstrated to be accurate, a custom stage with five degrees of freedom and a standard stereotactic frame. Once the system was built and customized, we employed well-established methods and procedures to characterize it, using simulations, measurements with an acoustic hydrophone and exposures in a tissue-mimicking phantom.
Simulations of the location of the focus were validated using the experiments in the hydrogel phantoms, where thermal lesions were generated and their position measured relative to that of the transducer. The disparity between the simulations and measured results was less than 1%. Regarding the simulations of the acoustic pressure at the focus, these were highest with the cone and inflated membrane. These results were not surprising seeing that this configuration ensured that there was degassed water in the entire beam path from the face of the transducer to the focus. In free field the acoustic pressure at the focus was only 19% of that with the inflated membrane, indicating the effect of having non-degassed water in the beam path, where acoustic cavitation can occur. Cavitating bubbles will not only absorb acoustic energy but under specific conditions, also reflect the transmitting beam. Both these factors will contribute to energy loss at the focus [21]. The acoustic pressure with the cone was similar (18%) to that in free field, indicating no effects of the presence of the cone. With a deflated membrane, the pressure was 30% lower than with the membrane inflated, further demonstrating the effects of non-degassed water, which was present at the focus.
Overall, the device design was found to be efficient for transmitting the ultrasound energy and localizing it at the focus. The results of the simulations were more or less in accordance with the hydrophone measurements for the cone with the membrane, both inflated and uninflated. Relatively larger disparities between measured and simulated values were observed for conditions without the cone or with the cone without the membrane. This was most likely due to the inability to accurately simulate the effects of the non-degassed water on the propagating beam compared to degassed water, where