Technical Principles and Clinical Workflow of Transcranial MR-Guided Focused Ultrasound Open Access

As of early 2020, over 50 institutions worldwide offer transcranial magnetic resonance-guided focused ultrasound (MRgFUS) as a less-invasive treatment option for thermoablation in the brain and >10 are involved in clinical testing of FUS combined with microbubbles for targeted drug delivery to the central nervous system (CNS) [1]. Other applications of FUS in the brain that are currently under clinical investigation include neuromodulation and ultrasound-assisted clot lysis (i.e., sonothrombolysis) [2, 3]. Owing to the appeal and potential benefits of less-invasive surgical techniques, MRgFUS thermoablation has superseded previously developed lesioning approaches at many institutions, particularly for unilateral thalamotomy in the treatment of medication-refractory essential tremor (ET) [4-6]. With demonstration of the cost-effectiveness of transcranial MRgFUS relative to standard surgical treatments for ET, this trend is poised to continue for other intracranial ablative indications [7, 8]. Additionally, low-intensity pulsed FUS and circulating microbubbles can increase blood-brain barrier (BBB) permeability to enable localized drug delivery to the CNS and is a burgeoning field of clinical research built on a promising line of preclinical investigations since the initial feasibility was demonstrated in 2001 [9-15]. If shown to be safe and effective in a clinical setting, this technology may open opportunities to treat diverse neurological conditions for which the BBB is a major impediment to therapeutic delivery. Other techniques to bypass the BBB exist for enhanced CNS drug delivery. However, some of these techniques have shortcomings and risks that limit more widespread adoption, such as the heterogeneous BBB opening (BBBO) and off-target toxicity associated with intra-arterial mannitol [16, 17].

Over the past few years, there has been increasing interest in the use of FUS as a tool for modulating BBB permeability in a number of clinical indications, ranging from cancer to neurodegenerative disease. This article provides an overview of the basic principles of transcranial MRgFUS, the current state of the technology, treatment protocols, and existing limitations. We emphasize the methodology and nuances of nonthermal microbubble-mediated FUS procedures (e.g., BBBO) and highlight areas of improvement for the treatment team.

The currently available clinical FUS brain devices can be stratified into MR-guided (ExAblate Neuro; InSightec, Israel), neuronavigation-guided (NaviFUS, Taiwan), and implanted (SonoCloud, CarThera, France) systems. Two ExAblate Neuro systems are available, and these systems share the same peripheral electronics but operate at different frequencies: a high-frequency (650–720 kHz) device for thermal ablation and a low-frequency (220–230 kHz) device for nonthermal microbubble-mediated treatments. The high-frequency ExAblate Neuro system is currently the only FUS device approved for thermoablation in the brain. The low-frequency ExAblate Neuro system, along with NaviFUS and SonoCloud, is the clinical-prototype device for FUS and microbubble-mediated BBBO. The ExAblate Neuro and NaviFUS systems use extracorporeal FUS transducers, whereas the SonoCloud device is an unfocused ultrasound transducer that is implanted in the skull cavity [18]. Several laboratory FUS brain systems exist and are at various stages of development and testing [19-21].

This article focuses mainly on the ExAblate Neuro devices, which have been employed for both thermal and nonthermal FUS brain treatments, since most of the clinical experience to date has been based on these devices. The ExAblate Neuro system consists of a 30-cm diameter hemispherical surface (“helmet”) that is lined with 1,024 independent transducer elements for ultrasound delivery. The helmet is fixed to a specialized MRI bed and is attached to the patient’s head using a stereotactic frame. The intervening space is filled with chilled degassed/deionized water, held in place by a rubber membrane fitted around the circumference of the head, which provides acoustic coupling and enables cooling of the scalp and skull during thermoablation procedures. Based on the prescribed target location, helmet orientation, skull properties (estimated from patient-specific CT images), intraoperative MRI data, and additional user input (e.g., no-pass zones for intracranial calcifications and water membrane folds), the MRgFUS system calculates the necessary amplitude and phase for each individual transducer element to focus ultrasound through the skull bone to the prescribed target(s) [22]. In general, only a subset of the array may be active for a given sonication as elements associated with no-pass zones or high incident angles are disabled [23]. For thermal treatments, the transducer amplitude distribution is often set to achieve a uniform acoustic intensity density on the outer skull surface [24] to minimize skull heating effects [25]. For nonthermal treatments, during which skull heating is less of a concern, the transducer amplitude distribution is often set to ensure each array element contributes equally to the acoustic intensity at the focal spot [24]. As discussed in further detail below, intraoperative MRI is carried out to provide real-time treatment guidance and dedicated single-element acoustic detector monitor for any cavitation activity generated during the FUS exposures.

A critical responsibility of the treatment team is to promote efficient propagation or delivery of ultrasound into the brain such that the intended biological effect(s) occur at the target(s). Propagation of ultrasound into a medium is dictated by the acoustic impedance, which is related to the material density and sound speed. In passing through the skull, ultrasound is reflected, attenuated, and transmitted. At the interface separating 2 distinct media, a higher mismatch in acoustic impedance (e.g., from scalp to bone) results in increased reflected energy and consequently decreased ultrasound transmission (Fig. 1). Temperature elevation is the primary result of attenuation in the skull bone. Attenuation varies as ultrasound penetrates through different tissue, indicated by an attenuation coefficient of 13.0 dB/cm for the skull and 0.6 dB/cm for the brain at 1 MHz [26]. For comparison, a 3-dB decrease translates to a 50% reduction in ultrasound intensity. In addition to attenuation, variations in skull thickness introduce aberrations to the ultrasound path, defocusing the ultrasound beam. Overall, ultrasound suffers from increased attenuation and distortion with rise in frequency [26].

Using a curved single-element transducer or a multielement phased array, focused ultrasound creates a tight high-pressure point with a sharp drop off outside the focus. Phased arrays permit electronic control over the beam geometry and direction, as well as provide the ability to mitigate skull-induced beam distortions, and thus provide increased flexibility relative to single-element transducers. The size of the focal region can be adjusted depending on the transducer design (e.g., geometry) and ultrasound parameters (e.g., operating frequency). The target brain region should ideally be spatially matched to the geometric focus of the transducer. Arrays can be mechanically positioned such that the focus is as close as possible to the target, after which the focus can be steered electronically through phasing within a smaller region. The low-frequency ExAblate Neuro system allows sonications within a sphere approximately 4 cm in diameter without repositioning the helmet. Overall, the treatment envelope is larger for the low-frequency versus high-frequency system, provided the helmet is repositioned to place the target within the treatment sphere. For high-frequency applications, treatment outside of these zones is suboptimal because of decreased efficiency of ultrasound delivery, increased skull heating, and potential off-target effects.

Thermal dose is an important concept in understanding FUS thermoablation applications in the brain. The biological effects resulting from thermal exposures depend on temperature as well as the exposure duration. For this reason, thermal dose was developed as a metric to evaluate the extent of tissue damage resulting from thermal exposures [27]. For a given exposure, thermal dose can be interpreted as the number of cumulative minutes at 43°C that would be required to achieve an equivalent biological effect. With 43°C chosen empirically from in vitro experiments, for every additional degree of temperature elevation, the time required for coagulative necrosis decreases by 50% [28]. In general, each target tissue has a different sensitivity to thermal dose though 240 cumulative equivalent minutes at 43°C (CEM₄₃) appears sufficient to induce necrosis in all tissue types [29]. Data from MRgFUS thalamotomy procedures suggest the dose threshold is lower (∼100 CEM₄₃) in the human thalamus [30, 31]. The bioheat transfer equation is often used to model heat transfer and was formulated by Dr. Harry Pennes [32], a psychiatrist in New York City, through experiments in the human forearm. Dr. Pennes introduced the importance of convective heat loss through perfusion as well as diffusion to surrounding structures, which explains how large Virchow-Robin spaces, ventricular structures, and large-caliber blood vessels might act as heat sinks.

Cavitation refers to the interaction of acoustic waves with gas- or vapor-filled cavities. Preformed microbubbles are injected intravenously to facilitate stable cavitation necessary to open the BBB [33] and are available as ultrasound contrast agents. With stable or noninertial cavitation, bubbles oscillate, resulting in mechanical forces on the vessel wall thought to be mainly responsible for altering the structural and functional components of the BBB (e.g., occludin and ZO-1, caveolae, and P-glycoprotein). The safe and effective BBBO in nearly every brain region has been studied in animal models and reviewed in depth previously [34]. The extent of BBBO is dependent on a number of factors: ultrasound parameters (e.g., peak negative pressure, burst length, burst repetition frequency, and exposure duration), microbubble dose (e.g., concentration, size distribution, and circulation time), and intrinsic tissue properties (e.g., vascular density). One commonly used commercial ultrasound contrast agent is Definity® (Lantheus Medical Imaging), which are inert gas perflutren-containing lipid microspheres with a main size distribution of 1.1–3.3 μm and a half-life of in vivo approximately 1.3 min. Capillaries in the brain have a diameter between approximately 6 and 10 μm [35]. Such microspheres have similar rheology as red blood cells. Therefore, differences in vessel properties, like vascular density and fragility, can alter the result of BBBO. In summary, the therapeutic FUS can be classified by frequency, intensity, and addition of microbubbles, with different applications dependent on the dominant biologic effect (Table 1).

There are several modifiable factors to increase patient tolerability and overall treatment success. The overarching goal is to minimize acoustic attenuation in the near field – between the transducer and the target – and maximize energy deposition at the intended focus. These factors will be discussed in chronological order of the clinical workflow (Fig. 2).

Patients must meet disease-specific criteria for the MRgFUS. Medical contraindications for MRgFUS procedures include uncorrected coagulopathy and the American Society of Anesthesiologists physical status classification 4 or higher. Presurgical anesthetic assessments are arranged for all patients. In addition, patients should be made aware during the consent process of the need for a full head shave and placement of a stereotactic frame. Patients with mild-to-moderate claustrophobia may still be able to tolerate MRgFUS procedures on a case-by-case basis. Coagulopathies should be corrected such that the international normalized ratio is <1.2. Platelet count should also be above 100,000/μL. Patients on anticoagulation or antiplatelet therapies can be asked to hold these medications for the procedures if the associated risks are low.

Skull morphology characteristics are important determinants of temperature rise during thermal MRgFUS brain treatments. The skull density ratio (SDR) is one such metric calculated by the ExAblate Neuro systems. SDR is calculated based on patient-specific CT scans of the head, specifically the ratio between the Hounsfield intensity of the trabecular and cortical bone layers averaged across the trajectories between the target and each active element in the transducer array [36]. Thus, SDR is not a fixed value for each individual and depends on the anatomical location of the target as well as the number of transducers for which an unobstructed path to the target exists. To ensure proper estimation of SDR using existing transcranial MRgFUS systems, helical CT should be acquired covering the entire head with a slice thickness of 1 mm or less, reconstructed with the appropriate convolution kernel depending on the CT brand (e.g., bone plus kernel for GE) and with no gantry tilt.

Patients with low SDR values are generally associated with low sonication efficiencies in lesional applications (i.e., ratio of focal temperature elevation to the applied acoustic energy) that limit the achievable focal temperatures and can preclude lesion formation. Intracranial calcifications, scalp lesions, and hyperostosis frontalis impair ultrasound transmission and have a higher propensity to heat. The presence of these characteristics is relative contraindication for thermoablation. These regions, along with sensitive neurovascular structures, can be blocked as “no-pass zones” using the MRgFUS system software, which turns off the number of active transducer elements to minimize energy deposition in these regions. Numerical simulations of transskull ultrasound propagation may provide the opportunity to investigate treatment feasibility preoperatively [37-39].

The proportion of patients unsuitable for MRgFUS thermoablation depends on a number of factors and changes in this quickly advancing field. Patients with SDR below 0.30 and or hyperostosis frontalis will be cautioned regarding the higher risk of lesioning failure but are not absolute excluded during the screening process. Indeed, some institutions exclude patients from MRgFUS thermoablation procedures based on SDR (e.g., SDR <0.40–0.45) [5, 36, 40]. However, recent advancements in transcranial ultrasound focusing capabilities using the clinical device have resulted in improved sonication efficiency, which may ultimately increase the proportion of patients that are “treatable” via MRgFUS as well as enable successful lesioning in previously failed patients [41].

A patient might be excluded from participation in FUS BBBO procedures for a variety of study criteria including disease-specific scores, contraindications to delivered therapies, MRI, MRI contrast, and ultrasound contrast agent (e.g., Definity®). Specifically, contraindications to the latter include right-to-left shunts and severe cardiopulmonary disease among others that can be found in the product monograph.

For nonthermal microbubble-mediated MRgFUS procedures, skull properties are less critical for patient selection. First, the lower array operating frequency reduces the impact of skull-induced beam distortions [42]. Second, because the time-averaged acoustic powers required for microbubble-mediated BBB permeabilization are several orders of magnitude lower than those required for thermoablation [11, 12], the applied acoustic power can be increased until an appropriate level of cavitation activity is induced at the desired target without risk of skull heating. Patient tolerability of prolonged MRI sessions remains an important consideration. However, during nonthermal procedures, symptom feedback is less critical for treatment guidance, and thus patients can be more deeply sedated with careful selection of anesthetic agents.

Patient preparation for both thermal and nonthermal MRgFUS procedures follows similar protocols. Of importance is a meticulously clean shave to avoid air bubbles trapped in the hair [43]. Shaving cream may be used to reduce skin irritation after procedure, but should be entirely wiped away prior to patient setup. During the assemblage of the MR compatible Cosman-Roberts-Wells (CRW) stereotactic frame (Integra®; UCHRA), attention should be paid to correct placement of the numbered posts. In general, placement of the stereotactic frame can be placed under local anesthetic only, and wherever possible, intravenous (IV) adjuncts should be avoided due to side effects like nausea and oversedation. Furthermore, the frame should be positioned as inferiorly as possible on the head to maximize the number of active transducer elements (Fig. 3).

In placing the frame, we generally start with the front 2 head ring screws advanced 3/4 of the way in, in an attempt to avoid the temporalis muscle. We do not use a torque wrench when tightening the screws but do administer 1–2 more turns after they are finger tight. Adequate fixation of the head is particularly important for thermoablation treatments, both for precise stereotactic targeting and accurate temperature monitoring as the MR thermometry is sensitive to motion.

Attention should be given to the location of the target. Ideally, the target is placed as close to the transducer array’s geometric focus as possible. The helmet can be translated and rotated mechanically with respect to the patient’s head; however, when extreme targets are attempted, the range of motion can be limited physically by the stereotactic frame or MRI table. For cerebellar targets, for instance, the frame should be placed as inferior as possible. Shorter MR-compatible disposable head ring screws (Integra®; DHRSS5) can be used when a significant helmet tilt is necessary to provide maximal clearance, provided the subject head size permits. Finally, expanders or “ear” adapters are available to offset the frame even more from the helmet. A final check of any hyperflexion or hyperextension of the neck when the frame is parallel to the floor is important, particularly for patients with pre-existing cervical spine disease.

The clinical protocol for MRgFUS thermoablation and lesion prediction has been described previously [44-46]. Accurate location of the temperature elevation on axial and sagittal MR thermometry with sublesional sonications is required. MRI monitoring during sonications is an important feature of MRgFUS. Planning scans are acquired as described previously. Intraoperative thermometry is critical both for spatial accuracy and estimation of accumulated thermal dose, relevant to guiding treatment endpoints [30, 31]. Targeting accuracy of the thermal lesion can be tested with online MR thermometry in 3 orientations during subablative temperature sonications. Subsequent corrections can be made depending on the orientation and discrepancy if needed. In the MR thermometry employed by the ExAblate Neuro system, temperature changes are measured based on phase differences between dynamic frames in a single 2D plane. These differences are linearly related to temperature change in relevant ranges but are sensitive to motion. T2-weighted imaging is also helpful with regard to guiding treatment for instance to confirm lesion formation and assess size and location [47].

Bouts lasting 40–50 s of sequentially higher-power FUS in the range of 500–900 W are then delivered alternating with clinical examinations. Clinical examinations assess for treatment effects and adverse events that might advise spatial adjustment of the target. In general, we aim for 2–3 bouts of temperature elevation above 54°C. A T2 sequence is performed after a group of sonications between 50 and 57°C or when the patient experiences good clinical improvement to assess treatment endpoint. We aim for a lesion measuring 4–5 mm in diameter on the intraoperative T2-weighted sequence. The lesion size 1 day after the procedure is expected to expand by 1–2 mm. Lesion size can be predicted by the accumulated thermal dose model, which accounts for additional factors such as heat diffusion to adjacent tissue and oblique focus [30, 31, 48].

Overall, the total time on the MRgFUS system for thermoablation ranges from 2 to 3 h. In terms of anesthesia, sedation and analgesia are critically important during thermal FUS procedures for scalp heating and pain [49]. Scalp pain from high-power FUS is transient and does not typically require further treatment postoperatively. Acetaminophen is a typically effective analgesic for mild postoperative pain associated with cold water or stereotactic frame. A few hours of observation and supportive care is typically sufficient prior to patient being ready for discharge.

For BBBO procedures, a three-way stopcock is added on the IV line such that it is accessible for the ultrasound contrast agent injections while the patient is in the MRI bore (Fig. 3). Microbubble contrast agent administration can be carried out using repeated boluses or using a continuous infusion. For bolus administrations of Definity® as an example, a 4 μL/kg dose is injected 30–40 s prior to the start of sonication followed by a 5- to 10-cc saline flush enough to clear the microbubbles from the IV line [33]. The half-life of Definity® is 1.3 min, allowing approximately 5 min of effective sonication, and must therefore be redosed every 5 min. While the maximum indicated dose of Definity® for diagnostic imaging is 20 μL/kg, up to 100 μL/kg has been tested in initial clinical safety testing. Early MRgFUS trials adhered to 20 μL/kg, which was expanded following accumulation of safety data. For infusion administrations, 1.3 cc of Definity® is mixed into a 500-cc preservative-free saline bag and infused at 10 cc/min. Infusion allows this aspect of the procedure to be operator independent and allows continuous sonications, saving time and frequent interactions with the IV line. The potential disadvantages of infusion include settling of microbubbles over time, agent waste, and a lower temporal-peak microbubble concentration compared to bolus administrations. Finally, administration of O₂ has been shown to reduce the effectiveness of BBBO in animals [50], potentially due to decreased concentration of microbubbles in cerebral circulation from vasoconstriction [51]. Therefore, ideally, patients are removed of supplementary oxygen during sonications if safe to do so.

Because microbubble-mediated FUS applications are affected by the concentration and size distribution of microbubbles, operators should also be well versed with appropriate handling of Definity® and other similar ultrasound contrast agents [52]. Definity® is stored long term at 2–8°C and does not contain bacterial preservative. The vial should be allowed to reach room temperature (∼20 min) prior to activation, as activation temperature has been found to affect the resulting size distribution [53]. Activation involves mixing in the VIALMIX® apparatus (Lantheus Medical Imaging) for 45 s immediately before use, whereby the liquid turns from clear to milky. The microbubbles settle with time, so should be resuspended with repeated inversions of the vial or syringe for 10 s. Only 18- to 20-G needles should be used to draw up the Definity®, without injecting air into the vial. Previous work has shown rapid injection with small needles can damage the microbubbles, resulting in considerable differences in concentration and size distribution [54].

During planning stages, target regions are hand drawn over the planning MRI. For BBBO, polygon tools can be used to place a treatment volume (TV) with spacing of sonication spots predefined (Fig. 4). Smaller grid spacings (e.g., 2.5 mm) are preferred for more homogeneous BBBO and smaller anatomic structures. Each subspot has a focal depth of approximately 7 mm, which is about the full width at half maximum along the 230-kHz ultrasound beam [19].

Each TV consists of 1 bout of sonication and estimates the overall cavitation dose delivered. The maximum number of spots for TV – approximately 30 – takes into account the distance between spots and ultrasound parameters (e.g., duty cycle and repetition frequency) that are permissible via electronic steering of the transducers. However, larger volumes or more spots often result in greater heterogeneity of cavitation dose, and any spot in the TV may terminate the sonication if a dose threshold is reached. Currently, the power is not modulated by spot acoustic feedback.

For BBBO, MR thermometry is less useful as previous studies have shown macroscopic temperature changes are negligible [11, 12]. T2*-weighted gradient echo and susceptibility-weighted imaging sequences are useful to evaluate for microhemorrhages. Additionally, acoustic emission data can be used to predict the induced biological effects [55, 56]. T1-weighted MRI after gadolinium injection can visualize changes in BBB permeability and help determine treatment endpoints. However, once gadolinium is administered, it remains in the brain for the duration of the procedure. Thus, contrast-enhanced MRI lacks fidelity as a continuous measure of the effect of ultrasound.

Alternatively, actively controlled FUS exposure using acoustic-based monitoring is an emerging strategy to automate safe and effective procedures [57-59]. The detection of acoustic signals by passive cavitation detectors within the helmet and real-time decomposition into a frequency spectrum can be used to assess characteristics of the microbubble activity in situ [60]. Further, the use of multielement hemispherical detector arrays allows three-dimensional spatial localization of the induced microbubble activity within the brain [61], which has been demonstrated in several preclinical studies [19, 62]. This approach has recently shown utility in predicting FUS bioeffect volume distributions [63]. Acoustic emission analysis offers high fidelity monitoring of the BBBO process, promising to render the procedure safer and more efficient and real-time MRI unnecessary. Work is ongoing to predict the quality of BBBO with the cavitation dose detected from the calibration of cavitation detectors on the current low-intensity MRgFUS clinical prototype.

There is no scalp pain during FUS BBBO procedures, rather participants complain of coldness from subroom temperature water circulation. Because BBBO procedures may be prolonged depending on the total TV, up to 4 h within the device, participants are prone to becoming anxious or bored due to long periods of immobility. Other participants develop claustrophobia over repeated procedures. We find dexmedetomidine infusion to be helpful in maintaining patient cooperation and comfort during these long procedures. Other anesthetic protocols used for awake craniotomies might also be useful. Warming blanket or other devices are helpful in improving the tolerability of the procedure and postoperative recovery.

Similar to thermoablation, postoperative recovery requirements are minimal. Postoperative physical exams are conducted to document adverse events such as those related to FUS or the procedure. Patients might require a brief period of recovery with supportive care (e.g., non-opioid analgesia and antiemetics) prior to being ready for discharge.

In general, structural T1- and T2-weighted sequences are acquired to assess size and morphology of lesions. These characteristics, along with spatial location, have been shown to be correlated with clinical outcomes as well as adverse events [64, 65]. Time course of lesion development involves expansion of the lesion in the initial 24 h, stabilization, and potential regression over the course of months [66, 67]. Early postoperative susceptibility-weighted imaging has been found to be predictive of lesion appearance at 6 months [66].

For BBBO procedures, MRI contrast agent extravasation on MRI can be used to measure BBB permeability and taken as a surrogate for drug delivery [68]. For instance, Gadovist® is approximately 600 Da and does not cross the intact BBB. The appearance of enhancement within the sonicated target can appear well demarcated, such as reported in a group of patients with AD [12], or heterogeneous and “fluffy” as found in a group of patients with glioblastoma [13], possibly depending on underlying tissue properties (e.g., interstitial pressure, grey vs. white matter, and vascular density). Blood-pool gadolinium-based contrast agents (e.g., gadofosveset) that bind to albumin (66 kDa) have been shown to enhance sonicated targets following BBBO as well but at substantially lower levels than nonblood pool agents [69]. Gadofosveset provides lower sensitivity of BBBO but may provide greater validity as a surrogate marker for enhanced delivery of larger molecular drugs.

The presence of MRI contrast agent within the subarachnoid and perivascular spaces appears highly intense on fluid-attenuated inversion recovery (FLAIR) sequences, which also has a differential diagnosis of acute subarachnoid hemorrhage or meningitis [70]. Contrast enhancement on FLAIR after BBBO allows visualization of gadolinium clearance through glymphatic pathways and resolves once BBB permeability is restored [71]. In addition, T2, T2* gradient echo, and diffusion weighted imaging are useful for assessing adverse events following sonications, such as swelling, bleeding, and ischemia. In our experience, post-BBBO T2* changes tend to resolve on follow-up studies the day after the procedure [12]. Functional imaging, diffusion tensor imaging, and arterial spin labeling techniques are amongst the array of research MR sequences available to further characterize the biological effects of FUS-mediated increases in BBB permeability [72, 73].

Following MRgFUS procedures, participants are admitted to the hospital for observation and often can meet standard criteria for discharge within a few hours. For thermoablation, the follow-up MRI 1 day after treatment provides a more accurate assessment of lesion size. A delayed clinical assessment here is important given new adverse events. For BBBO cases, MRI evaluation 1 day after treatment is useful for assessing restoration of BBB permeability and evolution of any signal change on other MR sequences.

Existing challenges in MRgFUS thermoablation include expanding the treatment envelope, increasing the number of eligible candidates for MRgFUS, streamlining the procedure workflow, and incorporating more patient-centered outcome measures. The treatment envelope refers to the volume of brain tissue within the skull cavity that can be targeted and treated successfully. With existing MRgFUS technology, sonication efficiency is higher for central targets (e.g., thalamotomy) than for peripheral targets (e.g., capsulotomy) and is higher in some patients than others [74]. One approach to improve sonication efficiency is to employ transskull ultrasound focusing based on acoustic measurements collected during treatment [75, 76]. A similar approach has been demonstrated to be successful and safe in patients with ET using the high-frequency ExAblate Neuro system [41]. In terms of retreatment or bilateral treatments, case series of patients with obsessive-compulsive disorder and major depressive disorder have demonstrated that bilateral MRgFUS capsulotomy is feasible [77, 78]. Clinical testing of bilateral staged thalamotomy and subthalamotomy is ongoing (NCT04501484, NCT03465761, and NCT-03964272). Data from early MRgFUS thalamotomy cases show sonication efficiency can decrease over time within a single session potentially caused by changes in acoustic properties of the skull and brain tissue due to heating [48]. This limitation will have implication on the feasibility of bilateral procedures conducted in a single session. In recent years, there has also been an increased engagement of the entire health care team in understanding the unique implications of this novel technology [49, 79].

The treatment envelope within the brain is substantially expanded for BBBO procedures, allowing, for instance, treatment of the hippocampus and medial temporal lobe [15]. However, it is still our experience that treating extreme locations within the skull (e.g., the base of the anterior middle fossa, occipital pole, and lateral cerebellum) is more challenging using MRgFUS brain systems. This is largely due to the difficulty of placing the stereotactic frame such that the helmet’s geometric center is as close as possible to the target. In this regard, a freely moving FUS array could offer greater flexibility in targeting. The development of a frameless system, such as, for example, with the NaviFUS system, where targeting is based on preoperative MRI, may facilitate this. Furthermore, frameless systems are generally better tolerated by patients and reduce procedure time. However, the greater flexibility comes with a tradeoff of a decrease in precision. The spatial error of frameless systems is at least 2 times greater than frame-based systems [80].

Automation of ultrasound exposure delivery and monitoring to accurately predict treatment endpoint will enable larger tissue volume to be treated in a shorter amount of time. Acoustic emission patterns from ultrasound-stimulated microbubbles have been shown to predict successful BBBO [58, 59, 81]. The same signals have been shown to correlate with the level of increase in BBB permeability, as measured permeability coefficient Ktrans from dynamic contrast-enhanced MRI [82]. This relationship, however, has yet to be studied in human subjects. Acoustic feedback, compared to feedback by contrast-enhanced T1w MRI, offers higher fidelity during continuous monitoring since gadolinium has a dose limitation and half-life. On the other hand, T2* acquisition during sonications can be useful for early detection of adverse events. Nevertheless, the signal-to-noise ratio of intraoperative MRI acquisitions still desires improvement. In early clinical MRgFUS BBBO studies, intraoperative MR thermometry confirmed that focal temperature elevations are negligible during the low duty-cycle exposures employed for this application [12].

Finally, the practicalities of combining FUS with a drug and the effect of FUS on its pharmacokinetic profile still have many unanswered questions. For example, what is the optimal time to inject the drug? Dynamic microscopy studies in animals show fast and slow leakage of intravenously injected tracer after FUS BBBO and peak increase in permeability within 5 min of sonications [83, 84]. Notably, fast leakage appeared to dominate with higher acoustic pressures and larger caliber (30–70 μm) vessels [84]. Together, these results suggest the drug should ideally reach peak plasma concentration during sonications. Will it be feasible to combine FUS to therapies with dosing regimen as frequently as weekly? By how much does FUS improve the tissue concentration of a drug and how should the systemic dose be adjusted as a result, if at all? Several animal studies have looked at this for specific drugs or more generally using Ktrans in different tissue including tumor [85, 86]. Another approach to addressing this question in vivo is through radiopharmaceuticals, which has been investigated in animal models [87]. However, radiopharmaceuticals are specific to each drug and costly and time consuming to develop.

This review provides an overview of how to optimize clinical research protocols for transcranial applications of MRgFUS, with an emphasis on microbubble-mediated BBBO procedures. Continued understanding of the associated technical parameters, together with further advances in MRgFUS technology, will help standardize and optimize treatment procedures, as well as expand indications in the future.

Not applicable.

N.L. has received honorarium for serving on an expert steering committee for the Focused Ultrasound Foundation, as well as research support from the Focused Ultrasound Foundation and InSightec. K.H. is an inventor of intellectual property related to the brain application of focused ultrasound owned by the Brigham and Women’s Hospital and Sunnybrook Research Institute. Other authors have no potential conflicts of interest to disclose.

Y.M. was supported by the research fellowships from the Natural Sciences and Engineering Research Council, Physicians’ Services Incorporated, and Hold’em for Life Foundations.

Y.M., R.M.J., B.D., Y.H. C.B.P., and S.S. were all involved in data collection, writing, and editing of the manuscript. C.H., N.L., and K.H. critically reviewed and approved the manuscript.