Future Prospects of Ultrasound: Role of Low‐Intensity Transcranial Ultrasound Stimulation in the Treatment of Neurological and Psychiatric Disorders

ABSTRACT

Background

Neurological and psychiatric disorders pose a persistent global health burden, with conventional therapies often providing inadequate relief. Transcranial ultrasound stimulation (TUS) has attracted considerable attention as a non‐invasive neuromodulation technique offering high spatial resolution and the distinctive capacity to modulate deep brain regions.

Methods

This review synthesizes recent basic and clinical research on TUS across neurological and psychiatric disorders, and consolidates emerging evidence on its mechanisms, therapeutic applications, and safety considerations.

Results

Emerging evidence suggests that TUS may drive sustained neural remodeling—including synaptic plasticity, neurogenesis, and anti‐neuroinflammatory effects—raising the possibility that TUS could eventually contribute to disease‐modifying strategies. However, disease modification by TUS has not yet been demonstrated in humans; the evidence supporting this possibility is derived predominantly from preclinical studies and remains a hypothesis requiring confirmation in large‐scale, controlled clinical trials. Building on these frameworks, we further propose a working hypothesis—termed mechanotransduction‐to‐oscillation encoding—which posits a nonlinear encoding relationship between the pulse repetition frequency of TUS and the intrinsic oscillatory properties of the stimulated region.

Conclusion

This hypothesis remains largely speculative and is offered as a conceptual framework to generate testable predictions and guide future investigations. Together, these frameworks provide a cohesive strategy for advancing TUS from an exploratory neuromodulation technique toward a potentially clinically translatable precision intervention, while acknowledging that significant technical and clinical hurdles remain.

1. Introduction

Neurological disorders affect the brain and spinal cord, whereas psychiatric disorders primarily disrupt mood, thought, and behavior. Neurological conditions often present with motor or sensory deficits, while psychiatric disorders more commonly feature emotional and behavioral disturbances, though both may involve cognitive impairment [1, 2]. Currently, treatment options for these disorders include pharmacotherapy, interventional or surgical treatment, psychotherapy, and physical therapy, among which pharmacotherapy is the most common treatment method [3]. However, current treatments have significant limitations [3]. Therefore, finding more effective treatment methods is a research priority for addressing neurological and psychiatric disorders.

Neuromodulation has garnered significant attention for their application in the treatment of neurological and psychiatric disorders. Neuromodulation refers to biomedical engineering techniques that use physical means to target specific regions of the nervous system, either promoting or inhibiting neuronal activities to treat or improve symptoms of diseases. The brain areas involved in neurological and psychiatric disorders do not only include the cerebral cortex, such as the motor cortex and dorsolateral prefrontal cortex (DLPFC), but also include deep brain regions such as the basal ganglia, thalamus, hypothalamus, hippocampus, and amygdala [4]. Non‐invasive neuromodulation techniques, such as transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS), have not only been employed in the treatment of depression using the DLPFC but also applied to the primary motor cortex (M1) and supplementary motor area (SMA) to manage motor symptoms of neurodegenerative disorders [5]. However, TMS and tDCS have limited stimulation depth and can only directly stimulate the cerebral cortex; their effects on subcortical regions rely on indirect modulation via cortical–subcortical functional connectivity, with tDCS also suffering from relatively low spatial resolution. Deep brain stimulation (DBS) could target subcortical area for Parkinson's disease, dystonia, essential tremor, treatment‐resistant depression, and other disorders. It is invasive and involves implantation of electrodes into the deep brain regions and is associated with surgical risks such as hemorrhage and infection [5].

Low‐intensity transcranial ultrasound stimulation (TUS) is a non‐invasive neuromodulation technique that uses low‐intensity pulsed ultrasound waves to penetrate the skull and regulate neural tissues [6]. It has high spatial resolution and deep penetration capability, which enables precise stimulation of the cortex and deep subcortical brain regions while minimizing the impact on surrounding neural tissues [6]. Therefore, TUS not only overcomes the limitations of stimulation depth and low resolution of traditional non‐invasive neuromodulation, including tDCS and TMS, but also avoids the surgical risks and complications associated with invasive neuromodulation such as DBS [6]. This paper summarizes recent clinical and basic research on the treatment of neurological and psychiatric disorders with TUS. Particular attention is paid to the potential of low‐intensity TUS to target specific regions of the cortex and deep brain regions.

Several recent systematic and narrative reviews have thoroughly examined TUS mechanisms, safety profiles, and clinical efficacy. The present review distinguishes itself from these prior contributions in two specific ways: (1) we synthesize emerging evidence that TUS may induce structural and functional neural remodeling including synaptic plasticity, neurogenesis, and network reorganization, raising the possibility of shifting the therapeutic paradigm from symptomatic relief toward potentially disease‐modifying interventions—although these effects have not yet been demonstrated in humans; (2) we adopt a disease‐specific organizational structure while critically evaluating evidence quality and highlighting inconsistencies, moving beyond a simple listing of studies toward a critical synthesis that identifies both converging evidence and research gaps.

2. Composition and Regulatory Mechanism of TUS Devices

2.1. Composition of TUS Devices

As shown in Figure 1, which was created with http://biogdp.com/ [7], the key components of TUS devices used in human studies include an ultrasonic generator, a focusing device, a control unit, ultrasonic transducers, and a monitoring system. Its basic mechanisms are related to the physical properties of ultrasound and its interaction with biological tissues [8].

FIGURE 1.

Schematic diagram of the key components of a TUS system for human use.

It is important to note that TUS systems differ considerably according to their intended use and implementation. A recent registry‐based analysis of global clinical trials categorized TUS devices into three main groups based on their technical implementation and clinical deployment context: (1) wearable systems, which are lightweight, head‐mounted devices designed for repeated or prolonged use in outpatient or home settings; (2) portable neuronavigation‐guided systems, which integrate optical tracking and individual brain imaging for precise targeting in clinical or research environments; and (3) MRI‐integrated systems, which are designed for use within the MRI scanner environment, enabling real‐time imaging feedback during stimulation [8]. More recently, a fourth category—conformable adhesive ultrasound patches—has emerged, consisting of flexible, skin‐attached devices that maintain stable acoustic coupling for extended periods and have been demonstrated in human applications [9]. This classification framework provides a practical overview of the currently available TUS technologies and highlights the diverse translational pathways for non‐invasive neuromodulation.

2.2. Mechanisms of TUS

The neuromodulatory effects of TUS are mediated by both mechanical and non‐mechanical mechanisms. Mechanical effects include acoustic streaming, radiation force, and cavitation (including stable and inertial cavitation), which generate localized pressure waves and microstreaming [8]. Non‐mechanical effects primarily involve mild thermal changes resulting from energy absorption by tissues [8]. While the thermal increase is deliberately maintained within a safe subthreshold range (≤ 1°C) to avoid injury, this slight warming may still modulate neuronal excitability, synaptic transmission, and cerebral blood flow, collectively contributing to the neuromodulatory response [10]. However, existing literature indicates that the dominant mechanism of low‐intensity TUS is mechanical rather than thermal effects, as the low‐intensity protocol itself is designed to minimize thermal effects while preserving the mechanical transduction pathway [6].

Recent studies have identified mechanically sensitive ion channels as the primary molecular sensors that convert ultrasound‐induced mechanical stress into neural activity. Several mechanosensitive channel families—including Piezo1, TRPC6, TRPC4/5, and K2P channels—have been implicated in TUS neuromodulation, although which predominates under specific conditions remains unclear. Piezo1 activation mediates calcium influx and neurotransmitter release, with knockout or inhibition attenuating TUS responses, yet a direct comparative study suggests Piezo1 is unlikely to be the sole mediator. TRPC6 knockout eliminates ultrasound‐induced neuronal activation, while TRPC4 and TRPC5 form a protein complex that facilitates the ultrasound response. K2P channels (dual‐pore domain potassium channels) have been confirmed as low‐threshold mechanosensitive effectors directly modulated by focused ultrasound. Other candidates include acid sensing ion channel 1 A (ASIC1a), cystic fibrosis transmembrane transduction regulator (CFTR), α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazole‐propionicacid receptor (AMPAR), and N‐methyl‐D‐aspartate receptor (NMDAR) receptors. Additionally, TUS can activate endogenous mechanosensitive proteins and engineered acousto‐genetic actuators (e.g., MscL‐G22S), enabling cell‐type‐specific modulation [6]. Overall, mechanical transduction via ion channels is a core driving mechanism, though the relative contribution of specific channel families likely depends on cell type, stimulation parameters, and experimental conditions.

In addition to immediate neural modulation, TUS can also induce sustained neuroplasticity changes, a process critical for treating neurodegenerative and psychiatric disorders. TUS promotes synaptic plasticity through a mechanism analogous to long‐term potentiation (LTP) [11]. In rodent models of Alzheimer's disease (AD), repeated TUS enhances hippocampal neurogenesis, elevates brain‐derived neurotrophic factor (BDNF) levels, and activates the tropomyosin receptor kinase B (TrkB) signaling pathway [12, 13]. Similarly, in a depression model induced by chronic unpredictable stress, TUS to the hippocampus and medial prefrontal cortex ameliorated depressive‐like behaviors, accompanied by elevated BDNF and restored monoamine levels [14]. In Parkinson's disease (PD) models, TUS enhances tyrosine hydroxylase (TH) expression, promotes dopaminergic neuron regeneration, and facilitates the release of glial‐derived neurotrophic factor (GDNF) and BDNF in the substantia nigra (SN) [15]. These effects collectively suggest that TUS has the potential to induce structural and functional remodeling, providing a biological rationale for exploring its possible disease‐modifying properties in future long‐term studies.

Another key mechanism of TUS is its regulation of neuroinflammation. Studies have demonstrated that TUS can suppress inflammatory responses in various neurological disorders [16]. In the epileptic mouse model, TUS reduces neuronal apoptosis and alleviates seizure severity by inhibiting the hippocampal c‐Jun N‐terminal kinase (JNK) signaling pathway [16]. In the MPTP‐induced PD mouse model, TUS targeting the subthalamic nucleus (STN) reduced microglial activation and markers of oxidative stress in the SN and striatum [17]. Chen et al. demonstrated that transcranial focused ultrasound can bidirectionally modulate epileptic seizures by altering extracellular glutamate and γ‐aminobutyric acid (GABA) levels in the epileptic focus, thereby highlighting the role of neurotransmitter balance in TUS‐mediated seizure control [18]. The anti‐neuroinflammatory effects of TUS also contribute to its neuroprotective role in AD and other neurodegenerative disorders [19].

Currently, multimodal neuroimaging provides critical insights into how TUS modulates brain activity at both the target level and broader functional network levels [20]. Functional magnetic resonance imaging (fMRI) studies have demonstrated that TUS targeting deep brain regions not only stably modulates blood oxygen level‐dependent (BOLD) signals in the stimulated areas but also influences their functional connectivity with cortical regions, suggesting that TUS activates distributed neural circuits rather than producing merely localized effects [20]. Similarly, in patients with impaired consciousness, TUS application to the central thalamus reduces BOLD signals in the frontal cortex and basal ganglia, and is associated with improved behavioral responsiveness [21]. Positron emission tomography (PET) further demonstrated that in patients with AD, TUS increased local glucose metabolic rates in the temporal lobe, cingulate cortex, and frontal cortex, and was associated with cognitive improvement [1]. In parallel with these imaging findings, electrophysiological recordings have provided direct evidence of TUS‐induced neural modulation. For example, when theta‐burst TUS (tbTUS) was applied to the internal globus pallidus, an increase in theta‐band power was observed in concurrently implanted DBS electrodes, providing direct electrophysiological evidence that TUS can modulate deep brain oscillatory activity [4]. Similarly, SEEG recordings in patients with epilepsy have shown that TUS reduces interictal epileptiform discharges, confirming its suppression of pathological neural excitability [22]. These neuroimaging and electrophysiological findings collectively support the concept of imaging‐ and physiology‐guided precision therapy: before treatment, structural and functional imaging is used to identify individualized targets; during treatment, stimulation parameters are monitored and adjusted in real time; and after treatment, imaging biomarkers are employed to evaluate therapeutic efficacy.

This multi‐scale organization progresses from fundamental biophysical forces to their molecular transduction mechanisms, from acute neuromodulation to sustained neuroplasticity, from synaptic and cellular events to the regulation of neuroinflammation, and ultimately ascends to the integration of effects at the whole‐brain network level as captured by multimodal neuroimaging. This hierarchy maps onto the causal chain through which TUS exerts its therapeutic effects—beginning with the physical interaction between ultrasound and tissue, proceeding through the activation of mechanosensitive ion channels and intracellular signaling, and culminating in circuit‐level reorganization that can be non‐invasively monitored in patients. By making these mechanistic layers explicit and linking each to specific disease‐relevant outcomes, it becomes possible to establish precise, quantifiable correlations between molecular and neurophysiological mechanisms and the observed clinical endpoints. This is not merely an academic exercise, but rather a prerequisite for advancing TUS toward mechanism‐based, individualized parameter optimization. Figure 2 briefly illustrates the mechanisms by which TUS ameliorates neurological and psychiatric disorders, which was created with Biogdp.com [7].

FIGURE 2.

The mechanisms by which TUS ameliorates neurological and psychiatric disorders.

3. TUS As Targeted Therapy for Neurological and Psychiatric Disorders

In the following sections, we review TUS applications across neurological and psychiatric disorders. Clinical studies are summarized in Table 1, and preclinical evidence is summarized in Table 2.

TABLE 1.

Clinical studies of TUS targeting different brain regions for neurological and psychiatric disorders.

Researchers	Year	Subjects	Stimulated area	Parameter	Results
Shin et al. [23]	2023	11 patients with CNP; age: 19~75 years old; disease duration: more than 6 months; they still presented obvious symptoms after more than 4 weeks of treatment; VPI ≥ 5; DPI ≥ 30; Skull density ≥ 0.4	ACC.	Frequency = 250 kHz; ISPTA < 0.72 W/cm2; TBD = 5~10 ms; DC = 50%~70%; total stimulation time ≤ 30 min; MI ≤ 1.9; 3 times/week; 2 weeks.	Pain in 11 patients was significantly reduced, and the effect lasted for more than 4 weeks. No side effects such as burns, headaches or epileptic seizures.
Lee et al. [22]	2022	6 patients with epilepsy; all of them had undergone SEEG implantation	Epileptic seizure area.	ISPTA < 2.8 W/cm2; PRF = 0.1 kHz; burst length/focus exposure = 3 ms; DC = 30%; 10 min/day; 3 days.	No BBB damage. TUS did not cause cortical or subcortical changes. No edema in the white matter. The frequency of seizures decreased in 2 patients and increased in 1 patient, but the seizures were subclinical. 4 patients experienced a decrease in IED frequency within 72 h of treatment. 1 patient had discomfort due to scalp heating. 1 patient had naming and memory impairment, but was not clear whether it was related to TUS. In 5 patients, their neural activities were slowed.
Stern et al. [24]	2021	8 patients with drug‐resistant and intractable temporal lobe epilepsy; age: 18~60 years; they were about to undergo temporal lobe resection.	The temporal lobe area to be resected	FF = 650 kHz; DC = 50% or 5%; SD = 0.5 or 30 s; PW = 2 or 0.5 ms; PRF = 0.25 or 0.1 kHz; duration: not described; 1 time.	None of the 7 subjects showed any damage. In one subject, eosinophilic neurons and blood extravasation were observed in the ultrasound‐stimulated area, but the significance of this phenomenon remains unclear. No significant changes in the neuropsychological tests after the surgery
Mahdavi et al. [25]	2023	25 patients with severe trGAD who met the DSM‐V diagnostic criteria.	Right amygdala.	FF = 650 kHz; PW = 5 ms; DC = 5%; PRF = 0.01 kHz; ISPPA = 14.30 W/cm2; ISPTA = 719.73 W/cm2; MI = 0.75; PNP = 0.61 MPa; alternate 30 s of treatment with 30 s rest; 10 min/session; Once a week for 8 weeks.	All patients tolerated TUS. No adverse reactions were reported. Symptoms improved in 16 patients.
Reznik et al. [2]	2020	24 patients with moderate depression; score of BDI‐II: 10~25.	Right frontotemporal region.	FF = 500 kHz; Sonication time = 30 s; PNP = 0.65 MPa; ISPTA = 0.071 W/cm2; ISPPA = 14 W/cm2; TUS 5 times within 7 days.	After 30 min of TUS intervention, their happiness increased. After 5 TUS sessions, anxiety improved.
Riis et al. [26]	2023	A 30‐year‐old woman with severe, treatment‐resistant non‐psychotic depression.	SCC	FF = 650 kHz; alternate 30 ms of treatment with 4 s rests; PNP = 1.0 MPa; ISPPA < 190 W/cm2; ISPTA < 0.72 W/cm2; continue for 2 min; for 10 times.	Depressive symptoms disappeared within 24 h of TUS and continued to remit for at least 44 days.
Sanguinetti et al. [27]	2020	25 healthy subjects	Right PFC.	FF = 500 kHz; PD = 65 μs; PRF = 40 Hz; DC = 0.26% or 0.5%; SD = 30 s or 2 min; duration: not described; 1 time.	TUS decreased resting‐state functional in resting state networks related to emotion and mood regulation. TUS induces positive emotions in volunteers.
Barksdale et al. [28]	2025	29 patients with MATRDs and 23 healthy subjects	Left amygdala	PRF = 0.01 kHz; DC = 5%; PW = 5 ms; ISPPA = 14.4 W/cm2; ISPTA = 719.91 W/cm2. Within each 10‐min interval, 10 on/off cycles of 30 s each are conducted; 5 times a week; 3 weeks.	TUS reduced oxygen‐dependent signals in the amygdala of both patients with MATRDs and healthy controls. TUS did not produce serious side effects. TUS improved the patients' anxiety and depression. TUS inhibited left and right amygdala activities.
Jeong et al. [1]	2022	10 possible AD patients; age: 65~85; MMSE < 20; CDR ≥ 1.	Left DLPFC.	FF = 250 kHz; TBD = 1 ms; PRF = 0.5 kHz; DC = 50%; duration = 20 min; MI = 0.6; 6 times in 2 or 3 weeks.	Regional glucose metabolic rates in the temporal lobe, cingulate gyrus, and frontal cortex significantly increased. Cognitive function, executive function, and memory improved. No damage in BBB.
Fuh et al. [29]	2025	90 years old; mild AD; clinical dementia rating of 0.5 or 1.0 and geriatric depression scale score was less than 8. The participants had been treated with acetylcholinesterase inhibitor for at least 6 months.	Temporal lobes	ISPTA = 0.5 W/cm2; DC = 20%; burst lengths = 2 ms; RP = 0.1 kHz; the TUS treatment was administered for a total duration of approximately 15 min per day, divided into three sessions of 5 min each, with a 5‐min rest period between each session; 6 weeks.	At 52 weeks, the change in ADAS‐cog score from baseline remained relatively stable in the TUS group compared to worsening in the placebo group, particularly in the memory domain. The change in MMSE score from baseline showed a significant benefit in the TUS group at 24 weeks compared to placebo, and this improvement persisted to 52 weeks. This study demonstrates the safety and feasibility of repeated TUS sessions in AD and suggests potential benefits in preserving cognitive function.
Samuel et al. [30]	2023	10 PD patients off medications.	Bilateral M1.	FF = 500 kHz; PRF = 0.005 kHz; SD = 200 ms; DC = 10%; p = 20 W; ISPPA = 2.26 W/cm2; ISPTA = 0.23 W/cm2; duration: each pulse lasts 80 s, for a total of 400 pulses; 3 times.	4 patients experienced temporary fatigue. None of the patients had side effects such as headache, neck pain, scalp heating, or muscle twitching. Excitability of the motor cortex significantly increased in all patients. Upper‐limb rigidity improved in 9 patients.
Vuong et al. [31]	2026	20 participants with PD, all of whom exhibited greater symptom severity on the right side.	Left ventral intermediate nucleus and left zona incerta.	FF = 500 kHz; PD = 0.5 ms; PRF = 0.1 kHz; DC = 5%; ISPPA = 6 W/cm2; stimulation duration, 30s; and inter‐stimulus interval of 30s (Figure 1B). The total stimulation duration was 10 min; 2 times.	TUS effects differed by target and tremor type. VIM‐TUS significantly reduced postural tremor but not rest tremor, whereas ZI‐TUS improved both postural and rest tremor. Although no overall group‐level rs‐FC changes were observed, individual Δrs‐FC of the ZI following ZI‐TUS correlated with tremor improvement, with similar findings for ΔfALFF.
Eraifej et al. [32]	2026	Inclusion criteria: diagnosed with PD; DBS targeted to STN; implanted with a percept implanted pulse generator. Have a measurable beta peak on screening. Exclusion criteria: unable to tolerate having their DBS turned off for the duration of the experiment. Have an extreme language barrier. Have a neurological or psychiatric comorbidity that may confound the results.	GPi	ultrasound was delivered in 800 ms pulse trains repeated every 2.5 s; PD = 90 μs; DC = 1.17%; PRF = 0.13 kHz; FF = 500 kHz; duration: 5 min; 2 times in 2 days.	Beta‐band power reduced in the ipsilateral STN. Beta power reduction was correlated between the ipsilateral, but not contralateral, STN and M1. Bradykinesia was also reduced.
Grippe et al. [33]	2024	20 moderately affected PD patients in on and off dopaminergic medication states (7 with and 13 without dyskinesia) and 17 age‐matched healthy controls in a case‐controlled study	Motor cortex	PRF = 500 kHz; DC = 10%; P = 20 W/cm2; ISPPA = 2.93 W/cm2; ISPTA = 0.29 W/cm2; duration = 80 s; 1 time.	The presence of dyskinesia did not affect tbTUS‐induced plasticity. TUS significantly increased MEP amplitude in controls and PD patients in dopaminergic medication states, but not in PD patients off medications. SICI was reduced in PD off medications compared to controls. TUS did not change in SICI or SICF. TUS improved motor function in PD patients. The presence of dyskinesia did not affect TUS‐induced plasticity.
Darmani et al. [4]	2025	10 patients with movement disorders (including PD and dystonia), and 15 healthy participants	2 sides globus pallidus internus	tbTUS: PRF = 500 kHz; duration = 120 s; pulse repetition interval = 200 ms; ISPPA = 30 W/cm2; SD = 20 ms. 10 Hz TUS: PRF = 500 kHz; continue for 40 s; Pulse repetition interval = 100 ms; ISPPA = 30 W/cm2; SD = 30 ms; duration = 40 s; 1 time.	tbTUS increases θ power, while 10 Hz TUS increases beta power. In healthy participants, tbTUS acting in the medial globus pallidus extended stop‐signal response time. Different parameters of TUS had different effects on pallidal activity.
Sarica et al. [34].	2026	17 patients with PD	M1, GPi, or the OCC	ISPPA = 5 ± 1.6 W/cm2 (OCC), 4.4 ± 1.2 W/cm2 (GPi), 3.2 ± 1.4 W/cm2 (M1); MI = 0.48 ± 0.08 (OCC), 0.45 ± 0.07 (GPi), 0.38 ± 0.08 (M1); PRF = 0.005 kHz; DC = 0.5%; duration = 120 s	TUS effects exhibited target specificity and brain state‐dependence. M1 stimulation significantly reduced STN beta oscillations and improved motor symptoms, whereas GPi stimulation increased beta oscillations without clinical benefits. Furthermore, the neuromodulatory effects of M1 stimulation varied between resting and movement states.
Mahoney et al. [3]	2023	1 females and 3 males with drug addiction	Bilateral NAc.	2 subjects were stimulated with low intensity TUS: p = 60 W; FF = 220 kHz; DC = 3.3%; alternate 5 s of treatment with 10 s rests; the pulse was continuously stimulated for 100 ms and stopped for 900 ms. Continue for 20 min, The other 2 subjects were stimulated with a stronger TUS: increased P to 90 W, and the other parameters were identical; All participants first received 10 min of sham stimulation (sham LIFU), followed by 20 min of active stimulation (active LIFU). The sham stimulation served as a control, during which the device delivered no energy.	No brain damage was observed by MRI. TUS alleviated craving for addictive substances for 90 days.
Cain et al. [21]	2022	11 acute DOC patients. Inclusion criteria: injury occurrence within less than 6 weeks, GCS score below 9, abnormal CT scan, prolonged loss of consciousness exceeding 24 h, and behavioral profiles indicative of coma or minimally conscious state, as assessed by the CRS‐Revised. Exclusion criteria involve deep sedation, pre‐injury history of neurological illness, and inability to safely enter the MRI environment.	Central thalamus.	PRF = 0.1 kHz; PW = 0.5 ms; CWF = 650 kHz; DC = 5%; ISPPA = 14.39 W/cm2 or ISPTA = 0.72 W/cm2; duration = 10 min; 8 patients underwent a single treatment session, while 3 patients underwent 2.	The patient's maximum response capacity was significantly enhanced, with a reduction in BOLD signal. No changes in vital signs (e.g., heart rate, blood pressure, or oxygen saturation) were observed during TUS administration.
Zhai et al. [11]	2023	26 schizophrenia inpatients with predominantly negative symptoms. ,13 patients in the active rTUS group and the other 13 patients in the sham group	DLPFC	FF = 500 kHz; ISPPA = 8.086 W/cm2; ISPTA = 0.404 W/cm2; PD = 500 μs; PRF = 0.1 kHz; ISI = 8 s; duration = 15 min; 15 times in 3 weeks.	TUS significantly alleviated negative symptoms and improved cognitive function in patients with schizophrenia, demonstrating a favorable safety profile. Furthermore, TUS intervention induced LTP‐like neuroplastic changes in the M1 of these patients.

Abbreviations: BDI‐II, Beck Depression Inventory‐II; BOLD, blood oxygenation level dependent; CDR, clinical dementia rating; CRS, Coma Recovery Scale; CWF, carrier wave frequency; DC, duty cycle; DSM‐V, diagnostic and statistical manual of mental disorders, fifth edition; FF, fundamental frequency; GCS, Glasgow Coma Scale; ISI, inter‐stimulus interval; ISPPA, intensity stimulated by pulsed plane wave; ISPTA, intensity stimulated by pulsed ultrasound; LID, levodopa‐induced dyskinesia; MDS, movement disorder society; MEFD, maximum exposure flow difference; MI, Mechanical Index; MMSE, mini‐mental state examination; OCC, occipital cortex; P, power; PDI, Pain Disability Index; PNP, peak negative pressure; PRF, pulsation rate frequency; PSC, primary sensory cortex; PW, pulse width; SD, sound duration; SEEG, stereo‐electroencephalography; SICF, short‐intracortical facilitation; SICI, short‐intracortical inhibition; TBD, time between bursts; TIc, thermal index of cranium; TIs, thermal index of soft tissue; VPI, Vagus Nerve Stimulation Pain Index.

TABLE 2.

Basic research on the treatment of neurological and psychiatric disorders by targeting different brain regions with TUS.

Researchers	Year	Subjects	Stimulated area	Parameter	Findings
Wang et al. [35]	2022	A CNP model of C57BL/6 mice established by chronic constriction injury surgery.	ACC (Bregma)	Intensity = 0.95 MPa; DC = 10%; PRF = 1.5 kHz; 15 min/day; continue to 21 days.	The number of c‐fos positive cells in ACC increased. MAP2, GAP43 and Tau protein levels in ACC neurons was reduced. Pain was reduced. TUS significantly and safely activates the ACC.
Kim et al. [36]	2024	Humanized transgenic homozygous Berkeley sickle mice (HbSSBERK)	S1	FF = 1500 kHz; BD = 200 μs; PRF = 0.040 kHz and 3 kHz; duration = 100 and 400 ms; ISoI of 2 and 4 s; total sonication time of 10 min, 20 min, and 1 h; 1 time.	significant change in Noxious heat hypersensitivity‐related and chronic hyperalgesia‐associated neural signals after focused ultrasound treatment
Zou et al. [16]	2021	A epilepsy model of C57BL/6 mice induced by KA.	CA1 and CA3 of the left hippocampus	DC = 50%; PRF = 0.010 kHz and 1 kHz; BD = 1 s; MI = 0.31; ISPTA = 0.046 W/cm2. 30 min/day; continue to 7 days.	Reduction of inflammation and apoptosis in the left hippocampus. Reduced activation of JNK signaling pathway in left hippocampus. Increased the number of c‐fos positive cells in the left hippocampus. Prolonged the incubation period of epilepsy.
Pang et al. [37]	2025	Penicillin‐induced epileptic mice	Hippocampus	FF = 1000 kHz; PRF = 0.5 kHz; DC = 50%; AP = 0.33 MPa; duration = 15 min; 1 time.	TUS significantly reduce the seizure firing rate.
Gao et al. [38]	2026	4‐AP injection in S1‐induced epileptic rats	Corpus callosum	PD = 20 ms; PRF = 0.001 kHz; DC = 2%; Pressure 127 kPa; duration: about 1 h; 1 time.	TUS can significantly reduce percent time in seizure, seizure duration, and seizure count, as well as increase inter‐seizure interval. They also observed a decrease in absolute power of the beta band and changes in the brain network, as evidenced by a decrease in synchronization and an improvement in interhemispheric balance.
Wu et al. [14]	2023	Depression model of C57BL/6 mice induced by corticosterone	ND	FF = 1100 kHz; PRF = 1 kHz; TBD = 0.5 ms; SD = 1 s; ISI = 1 s; DC = 50%; 30 min/day; 21 days.	TUS improved depression‐like behavior and anxiety‐like behavior. TUS reduced the concentration of NE in brain homogenate. TUS increased the level of 5‐HT in mouse brain homogenate. The number of c‐Fos positive cells increased.
Park et al. [39]	2021	6‐month‐old 5 × FAD mice	Entire brain	The 300 kHz ultrasonic beams pulsed at 0.04 kHz with 3‐ms pulse width were applied transcranially. The stimulation trains of 200‐ms ON and 800‐ms OFF periods were repeated in stimulation blocks of 10‐s ON and 30‐s OFF periods; 2 h/day; 2 weeks.	TUS reduced total Aβ42 levels and Aβ plaques in the hippocampus. The length and number of microglial processes decreased in pre‐ and infra‐limbic cortex and hippocampus
Yang et al. [40]	2024	APP/PS1 mice	Hippocampus	DC = 5%; RF = 0.001 kHz; Intensity = 0.3 MPa; ISPPA = 6 W/cm2; 15 min daily for a period of 28 days.	TUS can improve the learning and memory abilities of AD mice; reduce the phase‐amplitude coupling of delta‐epsilon, delta‐gamma and theta‐gamma frequency bands of local field potential, and increase the relative power of epsilon frequency bands in AD mice; reduce the spike firing rate of interneurons and inhibit the phase‐locked angle deflection between the theta frequency bands and the spikes of the two types of neurons that develops with the progression of the disease in AD mice.
Pandit et al. [12]	2019	AD model constructed by overexpression of K369I in K3 mice.	Whole brain	6 × 4 grid pattern; Pr = 0.65 MPa; PRF = 0.01 kHz; DC = 10%; 6 s/site; duration: not described; once a week for 15 weeks.	Reduced Tau protein phosphorylation. Reduced neurofibrillary tangle. Tau protein was cleared through activation of the autophagy pathway. Motor function and memory improved.
Xhima et al. [13]	2020	TgCRND8 mice	Basal forebrain	FF = 1680 kHz; BL = 10 ms; BRF = 0.001 kHz; continue for 120 s; 1 time.	Cholinergic neurons were activated. Concentration of BDNF increased. TrkA pathway was up‐regulated. Motor function improved. Phosphorylation of Tau protein was inhibited. The level of Aβ decreased.
Zhou et al. [17]	2021	PD model of C57BL/6 mice induced by MPTP	STN	FF = 3800 kHz; DC = 50%; PRF = 1 kHz; NTB = 1000; ISI = 4 s; Ispta = 0.43 mW/cm2; 30 min/day; Once a day, continued until the end of the modeling period.	Expression of c‐Fos in STN increased. Improve the motor ability Reduced the degeneration of striatum. Reduced inflammatory response in STN and striatum. Oxidative stress in the brain was reduced. Reduced the accumulation of α‐syn.
Wang et al. [15]	2019	PD model of C57BL/6 mice induced by MPTP	STN	Maximum pressure = 0.39 MPa; ISPPA = 5.1 W/cm2; FF = 500 kHz; PRF = 1 kHz; SD = 50 ms; DC = 5%; duration = 5 min; 1 time.	After ultrasound stimulation, mean power intensity in the beta band significantly decreased, and the phase‐amplitude coupling strength between the beta and high gamma bands and between the beta and ripple bands also became significantly weaker.
Xu et al. [41].	2020	PD model of C57BL/6 mice induced by MPTP	CS	FF = 1000 kHz; ISPTA = 0.1, 0.2, 0.3 W/cm2; 5 min/dar for 1 day or 10 min/dar for 2 days or 15 min/dar for 3 days.	DA levels in the corpus striatum increased. Motor function improved. Increased the number of TH positive neurons. Promoted the regeneration of neurons. Improve the permeability of neuronal cell membrane.
Sung et al. [42]	2022	PD model in rats induced by 6‐OHDA	RSN	FF = 1000 kHz; BL = 50 ms; DC = 5%; PRF = 0.001 kHz; ISPTA = 0.528 W/cm2; duration = 15 min; 5 days per week for a period of 6 weeks.	Motor functions improved. Reduction of TH levels in the SN and striatum; Increased levels of GDNF and BDNF in the SN. Reduced inflammatory response in the SN.

Abbreviations: AP, acoustic pressure; BD, burst duration; BL, beam length; CS, corpus striatum; ISoI, intersonication interval; MEP, motor evoked potential; MUP, maximum ultrasonic pressure; ND, nuclei dorsolateralis; RF, repetition frequency; RSN, right substantia nigra; SN, substantia nigra.

3.1. Neurological Disorders

3.1.1. Chronic Neuropathic Pain (CNP)

The anterior cingulate cortex (ACC) is a key node in the pain matrix and an emerging TUS target [35]. In a mouse model, TUS targeting the ACC safely suppressed pain by inhibiting local neural plasticity, though only morphological changes were examined [35]. More recently, targeting the primary somatosensory cortex (S1) with low‐intensity transcranial focused ultrasound significantly attenuated heat pain sensitivity in wild‐type mice and modulated hyperalgesia in a humanized mouse model of sickle cell disease [36], further supporting the potential of TUS to modulate pain processing at multiple nodes of the pain matrix.

Clinically, the consistent analgesic effects across diverse targets—ACC, posterior insula—suggest that TUS may act through distributed circuit modulation rather than a single pathway, which could be advantageous for treating the heterogeneous nature of chronic pain [23, 43]. However, most clinical evidence remains preliminary: only one pilot study has targeted the ACC in patients [23], and while the posterior insula findings in healthy volunteers are robust, translation to clinical pain populations is still lacking [43]. Furthermore, the optimal target for different pain subtypes and the stimulation parameters best suited for each target remain unknown. Future research should prioritize direct target comparisons within the same patient population and investigate whether target selection should be guided by individual pain phenotypes or neuroimaging‐based biomarkers.

3.1.2. Epilepsy

In a kainic acid‐induced mouse model, TUS applied to the left hippocampus reduced neuroinflammation and apoptosis, thereby decreasing seizure frequency [16]. Further research has confirmed that TUS can alter extracellular glutamate and GABA levels at the seizure focus, enabling both inhibition and facilitation of epileptic seizures depending on stimulation parameters [18]. Mechanistically, TUS targeting the motor cortex suppresses seizures through modulation of hippocampus‐cortical circuit activity, as demonstrated by reduced cortical power, weakened functional connectivity, and faster ictal‐to‐interictal transitions in awake epileptic animals [16]. Recent studies have also demonstrated that TUS targeting white matter can inhibit seizures in rat models [38]. Moreover, a closed‐loop TUS system guided by deep learning has been shown to effectively suppress epileptic seizures in mice, highlighting the potential of intelligent, real‐time adaptive stimulation strategies [37].

Clinical findings from small open‐label studies demonstrate that TUS can acutely reduce interictal epileptiform discharges (IEDs) and seizure frequency [24]. It is worth noting that one study observed eosinophilic neurons and blood extravasation in a single subject, the significance of which remains unclear and highlights the need for standardized histopathological assessment [24]. However, the maximum intensity used in that study (ISPTA = 5.76 W/cm²) substantially exceeds the FDA diagnostic ultrasound limit (ISPTA.3 = 0.72 W/cm²) and should be interpreted as a high‐dose exploratory exposure rather than a benchmark for low‐intensity TUS safety [44]. Several directions warrant particular attention in future research. First, the optimal stimulation parameters for seizure suppression—particularly the balance between inhibitory and facilitatory effects—should be systematically investigated to establish standardized protocols. Second, longitudinal studies are needed to determine the durability of TUS effects beyond the acute treatment period. Third, while current evidence is limited to drug‐resistant patients, future trials should explore whether TUS could be effective as an earlier intervention. Ultimately, parameter optimization in animal models should inform the design of larger, sham‐controlled trials with extended follow‐up periods to establish both efficacy and long‐term safety.

3.1.3. AD

Animal studies have shown that TUS can alleviate signs of AD by reducing the amount of amyloid‐β (Aβ) and Tau proteins in the brain and improve brain connectivity in AD mice [12]. Further research showed that the effects of TUS in AD are linked to hippocampal neurogenesis, BDNF increase, and reduced neuroinflammation [13]. Similarly, TUS treatment in APP/PS1 mice enhanced learning and memory performance and modulated neural oscillations [40]. In addition, TUS pulsed at 40 Hz reduced Aβ plaque burden and modulated brain rhythms in 5 × FAD mice, further supporting the potential of frequency‐specific TUS protocols in AD [39]. In brief, animal studies consistently demonstrate that TUS reduces effects that directly address core pathophysiological mechanisms.

Unlike the extensive volume of basic research, current translational studies on TUS in AD reveal a significant gap between robust preclinical findings and limited clinical evidence. In patients, the first randomized sham‐controlled trial provides encouraging evidence of cognitive preservation, while an open‐label PET study suggests modulation of brain metabolism in disease‐relevant regions, although the lack of sham control limits causal attribution [1]. The optimal stimulation protocol—including frequency, intensity, treatment duration, and focal versus whole‐brain targeting—remains undefined, and the heterogeneity of AD pathology further complicates target selection. Meanwhile, the patients' low level of cooperation represents one of the numerous challenges encountered. Future research should build on the positive signals from the randomized trial by conducting larger, multicenter, sham‐controlled trials with biomarker endpoints to determine whether TUS can truly slow disease progression.

3.1.4. PD

In 1‐methyl‐4‐phenyl‐1,2,3,6‐tetrahydropyridine (MPTP) PD mouse models, TUS of the STN alleviated neuroinflammation and improved motor function [17]. TUS of the striatum or SN increased dopamine (DA) concentration and TH activity [17, 41]. In addition, in a rat PD model established using 6‐hydroxydopamine (6‐OHDA), TUS promoted the diffusion of GDNF and BDNF in various brain regions [42]. Furthermore, TUS of the M1 significantly decreased parkinsonian‐related local field potential activity in MPTP‐treated mice [15].

In clinical research, tbTUS targeting the M1 increased cortical excitability [30, 33]. In addition, tbTUS of the internal globus pallidus (GPi) in the basal ganglia increased theta (4–10 Hz) frequency power in the GPi, suggesting potential therapeutic effects in PD patients [4], but M1 stimulation showed no significant clinical effect [45]. This suggests that direct modulation of the GPi, a central hub in PD motor circuitry, may be more effective than cortical stimulation. In a recent clinical trial comparing different subcortical targets, zona incerta (ZI) stimulation was found to be superior to ventral intermediate nucleus (VIM) stimulation for the suppression of rest tremor in PD patients [31]. A proof‐of‐concept study provided the first direct evidence that TUS can suppress pathological beta oscillations in the STN—a key electrophysiological biomarker of PD severity—although this finding was obtained in a small cohort of four patients and awaits replication in larger samples [32]. This demonstration of direct modulation of a pathophysiological biomarker represents an important step toward mechanism‐based TUS therapy. A recent study applied the TUS approach in 17 PD patients and demonstrated that TUS effects are both target‐ and frequency‐band specific: M1 stimulation reduced STN beta oscillation activity and improved motor signs, whereas GPi stimulation increased beta activity without motor improvement [34]. These findings provide mechanistic evidence that TUS can safely and selectively modulate pathological brain rhythms in the STN, supporting its therapeutic potential.

The therapeutic potential of TUS for PD has been explored through a wide range of targets, spanning cortical, striatal, pallidal, thalamic, and subthalamic regions in both animal models and human studies. Taken together, these clinical findings reveal several important patterns. First, deep targets that constitute key nodes within the basal ganglia–thalamocortical motor circuit—particularly the GPi, STN, and ZI—have shown more consistent clinical or electrophysiological effects than cortical targets such as M1, likely because they directly engage the circuit nodes most profoundly altered in PD pathophysiology. Second, the differential efficacy between ZI and VIM for tremor suppression highlights that even anatomically adjacent targets can produce distinct therapeutic profiles, possibly reflecting their divergent connectivity within motor and non‐motor circuits. Third, the successful modulation of pathological beta oscillations by STN‐targeted TUS is conceptually important because it demonstrates that a non‐invasive technique can directly engage a validated electrophysiological biomarker of disease severity, thereby providing a rational basis for parameter optimization. The recent expansion of this approach to a larger cohort further strengthens this paradigm by demonstrating that TUS effects are not monolithic but rather depend on both the targeted structure and the stimulation frequency [34]. However, direct comparisons between studies are complicated by substantial heterogeneity in stimulation parameters, outcome measures, and patient populations—ranging from off‐medication states to varying disease durations. Future work should prioritize head‐to‐head comparisons of target efficacy within the same patient cohort, as well as systematic investigations of how stimulation parameters interact with target‐specific circuit properties to determine clinical outcomes. Ultimately, the PD literature illustrates both the promise and the challenge of TUS: the availability of multiple candidate targets offers flexibility, but it also demands a mechanistic understanding of target–parameter interactions to move beyond empirical target selection toward rational, physiology‐guided neuromodulation.

3.2. Psychiatric Disorders

3.2.1. Depression

In a rat model of depression, TUS applied to the hippocampus and specific medial prefrontal subregions (prelimbic and infralimbic cortices) alleviated depressive‐like behaviors, and this therapeutic effect was associated with changes in the amount and distribution of BDNF in the brain [46]. In corticosterone‐treated mice, TUS of the dorsal lateral nucleus increased hippocampal BDNF and normalized the monoamine abnormalities—restoring diminished serotonin and countering the rise in norepinephrine [14].

The clinical evidence for TUS in depression, though still nascent, includes the first randomized, double‐blind, sham‐controlled trial of any TUS application in a psychiatric disorder. For example, five sessions of right frontotemporal TUS improved well‐being and reduced anxiety [2], while inhibitory TUS over the subcallosal cingulate cortex (SCC) produced rapid and durable remission in three treatment‐resistant cases [26]. A subsequent randomized, double‐blinded, sham‐controlled trial, SCC‐targeted TUS relieved depressive symptoms in 22 patients with depression [47]. In addition, TUS of the right inferior frontal gyrus in healthy subjects was found to regulate functional connectivity of emotion‐related neural networks and induced positive emotions [27]. It is worth noting that the evidence remains highly heterogeneous: the case series and open‐label studies provide preliminary signals but cannot establish efficacy, and the single randomized controlled trial requires replication in larger, more diverse patient populations. Furthermore, the optimal target remains debated—SCC, frontotemporal regions, and DLPFC have all shown signals of efficacy in different study designs, and it is unclear whether different targets correspond to different symptom dimensions, such as mood, anxiety, and anhedonia [2, 26, 27, 47]. Future work should systematically compare targets within the same patient cohort, investigate whether target selection should be guided by individual functional connectivity profiles, and determine whether rapid antidepressant effects are durable with repeated sessions. Mechanistically, the roles of local SCC modulation versus broader network‐level changes require clarification through concurrent neuroimaging in patients, while the molecular and cellular pathways underlying TUS‐induced antidepressant effects warrant further translational investigation.

3.2.2. Anxiety Disorders

In an exploratory open‐label trial, 25 participants with treatment‐refractory generalized anxiety disorder (trGAD) received 10 min of amygdala‐targeted TUS once weekly for 8 weeks [25]. The results showed that the TUS treatment reduced anxiety, and 16 patients showed significant clinical improvement. These results suggested the potential clinical effect of TUS in the treatment of trGAD. Another study using amygdala‐targeted TUS significantly alleviated mood, anxiety, and trauma‐related disorders (MATRD) [28]. Notably, neither of these studies included an independent sham or placebo control group for the therapeutic outcome assessment. The trGAD study was an open‐label single‐arm trial, and while the MATRD study employed a within‐subject sham control for target engagement, its treatment phase was also open‐label and lacked a parallel control group; therefore, the reported clinical improvements should be interpreted with caution and await confirmation in randomized controlled trials. Beyond the methodological limitations, the amygdala‐targeting approach is conceptually appealing given the amygdala's central role in fear and anxiety processing [48]. The consistent reduction in amygdala BOLD signals observed in the MATRD study provides initial evidence of target engagement, supporting the biological plausibility of this approach [28]. However, several open questions remain. First, the optimal stimulation parameters for amygdala modulation, particularly whether the parameters used in these studies represent the most effective or safe settings, require systematic investigation. Second, the durability of the observed clinical effects beyond the treatment period has not been established. Third, whether amygdala‐targeted TUS is broadly effective across different anxiety disorder subtypes, or whether its efficacy varies depending on symptom profiles or comorbidities, remains unknown. Given the high prevalence and disability associated with anxiety disorders, the promising preliminary signals from these studies warrant expedited follow‐up with rigorous sham‐controlled, parallel‐group trials that include objective biomarkers of target engagement and longer‐term follow‐up.

3.2.3. Substance Use Disorder (SUD)

The NAc is a mechanistically rational target for SUD given its central role in reward processing and craving [3], and the first‐in‐human pilot study provides initial evidence that TUS can safely modulate cue‐induced craving in humans [3]. However, this evidence must be interpreted with extreme caution. The sample size was very small, the study was open‐label, and the stimulation parameters varied across participants [3]. More importantly, a subsequent report of an adverse event in a similar protocol has raised critical safety concerns—although it appears that the acoustic parameters in that case exceeded established safety thresholds, this incident underscores the importance of rigorous parameter selection and monitoring in future SUD trials [44]. Given the vulnerability of the SUD population and the potential for high expectations to influence self‐reported craving outcomes, future studies must prioritize randomized, sham‐controlled designs with objective outcome measures (e.g., drug abstinence verified by toxicology, behavioral tasks of craving) and comprehensive safety monitoring. If these can be achieved, TUS targeting the NAc could represent a novel, non‐invasive approach to a disorder with limited current treatment options.

3.2.4. Schizophrenia

Zhai et al. applied low‐intensity repetitive TUS (rTUS) to the left DLPFC in schizophrenia patients predominantly presenting with negative symptoms. After 15 sessions, the active group showed significant improvements on the Scale for the Assessment of Negative Symptoms (SANS) and the Continuous Performance Test (CPT) [11]. This single randomized sham‐controlled study provides important proof‐of‐concept evidence that TUS may be safe and potentially beneficial for the negative symptoms of schizophrenia—an area with significant unmet therapeutic need [11]. However, this finding should be considered preliminary for several reasons. First, this is the only published study to date, and replication in independent samples is essential. Second, similar to other diseases, the optimal stimulation parameters have not been systematically explored, which indicates the parameters used in this single study may not be optimal. Third, the mechanism by which DLPFC‐targeted TUS improves negative symptoms—whether through direct modulation of prefrontal excitability, changes in dopamine release, or broader network effects—remains unknown. Given the chronic and disabling nature of negative symptoms and their poor responsiveness to existing antipsychotics [11], further research in this area is warranted. Future trials should include larger sample sizes, longer follow‐up periods, and mechanistic assessments (e.g., neuroimaging or electrophysiological biomarkers) to elucidate the neural basis of the observed clinical effects.

4. Challenges and Solutions for Clinical Application of TUS

4.1. Skull Attenuation and Focusing of Ultrasound

The skull impedes the transmission of ultrasound because of its high density. As a result, the energy reaching the targeted regions, especially for targets located in deep brain, is significantly attenuated. Lower‐frequency ultrasound waves have less attenuation by the skull, while higher‐frequency ultrasound waves provide better focality (though at the cost of greater attenuation by the skull) [10]. In addition, multi‐channel focusing technology can improve the focusing effect of ultrasound to enable greater concentration of energy at the target and reduce skull interference [10].

4.2. The Importance of Precise Positioning of Ultrasound Transducer

The selection of stimulation sites is a crucial factor in whether TUS can exert its therapeutic effects and prevent side effects. The compatibility of TUS with multimodal neuroimaging further advances the development of focused precision, which provides new possibilities for the precise treatment of neurological and psychiatric disorders. For instance, investigators can obtain individualized brain structural MRI data and then design a specific focusing plan to ensure that the TUS can accurately reach the target regions. Moreover, real‐time monitoring methods may allow clinicians to assess the mechanical effects of TUS in vivo and adjust stimulation parameters dynamically—thereby enhancing treatment precision and safety. Real‐time estimation of the TUS intracranial sound field can be achieved by constructing a deep learning surrogate model. This approach enhances the precision of focal positioning and pressure estimation through a dual‐module design.

4.3. The Prospect of Individualized Treatment

Patients with neurological and psychiatric disorders have variable symptoms and pathophysiological characteristics due to different environmental and genetic factors. Therapeutic regimen of TUS should be personalized according to the specific symptoms, patient characteristics, and treatment response of each patient to maximize the therapeutic effect and to avoid side effects. TUS treatment parameters may potentially be predicted and optimized by analyzing large patient datasets. With the development of artificial intelligence and big data technology, it may help to achieve more precise individualized TUS treatment plans.

4.4. Development of Portable TUS Devices

Portable treatment devices allow patients to receive treatment anytime and anywhere, thereby making the treatment more widely available and potentially enhancing the treatment effect. With further development of integrated circuits and low‐power chips, the energy consumption of the devices can be significantly reduced, which increases the feasibility for the development of portable TUS devices. Notably, early explorations in animal models are informative but require careful translation to humans, because the significantly greater thickness and higher density of the human skull attenuate ultrasound transmission far more than in rodents. Therefore, portable TUS devices designed for humans must address skull‐induced attenuation through strategies such as optimized energy delivery (while remaining within established safety limits), multi‐channel focusing, or personalized aberration‐correction strategies such as metamaterials.

4.5. Parameter Optimization

The establishment of standardized TUS stimulation protocols remains a critical bottleneck, requiring the integration of multidisciplinary knowledge spanning acoustics, neurobiology, and clinical medicine. Concurrently, the lack of elucidation of therapeutic mechanisms, particularly how molecular mechanisms drive clinical efficacy, limits its translational applications. Furthermore, the challenge of transitioning from a general protocol to an individualized, precise treatment plan represents a core hurdle for the next phase.

A complementary challenge is the limited understanding of how specific TUS parameters map onto clinical outcomes. A recent systematic review and meta‐analysis of human TUS studies explored the relationship between stimulation parameters and the directionality of neuromodulatory effects [49]. Although univariate analyses identified only weak statistical trends, the authors established an open‐access database to facilitate future parameter‐outcome analyses [49]. This initiative underscores the urgent need for standardized parameter reporting and large‐scale data aggregation to enable evidence‐based parameter optimization, a prerequisite for closing the gap between generic protocols and individualized precision TUS.

In the future, artificial intelligence analysis of multidimensional data will be essential to integrate individual anatomical structures, target molecular characteristics, and disease states, achieving dynamic optimization of treatment parameters, thereby bridging the gap from standardization to individualization.

5. Potential Risks in TUS

In clinical studies, the short‐term application of TUS generally exhibits good tolerance; however, it is imperative to acknowledge its potential biophysical risks.

5.1. Regulatory and Consensus Safety Guidelines

A critical distinction must be made regarding the origin of current safety limits. The food and drug administration (FDA) safety thresholds commonly referenced in the field—specifically, the acoustic output limits for MI and thermal index—were originally established for diagnostic ultrasound applications and are not directly applicable to therapeutic TUS interventions [50]. At present, there are no universally accepted safety thresholds specifically designed for therapeutic TUS. The ITRUSST consensus guidelines indicate that the use of ultrasound parameters within the diagnostic safety limits is generally considered safe, whereas the safety of exposures exceeding these thresholds remains uncertain and should be evaluated empirically in appropriately designed studies rather than assumed to be harmful or safe a priori. Therefore, each TUS protocol must be assessed on a case‐by‐case basis, with careful attention to the specific parameters, target depth, and individual patient characteristics.

5.2. Patient‐Specific Factors Influencing Safety

Importantly, ultrasound exposure parameters alone do not fully determine safety. Patient‐specific factors may substantially influence the risk profile, including metabolic status, underlying pathology, and the presence of intracranial air bubbles (e.g., following recent neurosurgical procedures) or calcifications, which may alter acoustic propagation and increase the likelihood of adverse effects. These anatomical and pathological variants should be systematically assessed prior to TUS administration to mitigate potential risks.

5.3. Reported Adverse Events and Evidence Gaps

A recent study reported an adverse event in which a patient experienced severe consequences following FUS treatment directed at the NAc [44]. This case has generated considerable discussion within the field, with particular attention to the transparency of acoustic parameter reporting and whether the applied stimulation parameters remained within established low‐intensity safety limits. A subsequent open letter from multiple international experts raised concerns about the non‐disclosure of fundamental acoustic metrics such as the Mechanical Index and questioned the study's adherence to the safety thresholds proposed by the ITRUSST consensus [44]. Critically, the acoustic parameters used in this case—particularly the Mechanical Index—appear to have exceeded the safety thresholds recommended for low‐intensity TUS. Therefore, this adverse event should not be interpreted as reflecting an inherent or unpredictable risk of standard low‐intensity TUS protocols that operate within established safety boundaries. Instead, it underscores the critical importance of transparent parameter reporting and strict adherence to safety guidelines to ensure patient safety in future studies.

In addition, a methodological caveat deserves acknowledgment. While some studies have reported no histopathological evidence of tissue injury following TUS, the same study also described the presence of eosinophilic neurons and blood extravasation in one subject, the significance of which remains unclear [24]. This observation highlights the need for standardized histopathological assessment protocols and rigorous safety reporting in future studies, as even apparently minor findings may have implications for long‐term safety. The classification of adverse events using a structured framework would facilitate systematic safety monitoring and cross‐study comparison.

5.4. Biophysical Mechanisms of Potential Risk

Beyond thermal effects, several biophysical mechanisms warrant consideration. Inertial cavitation—the rapid collapse of gas bubbles in the acoustic field—can produce localized mechanical stress and potential tissue damage, and its occurrence is influenced by pulse parameters and the presence of gas nuclei in the tissue. Standing waves, which can arise from reflection at the skull‐brain interface, may produce unpredictable hotspots of acoustic energy deposition, potentially exceeding intended intensities at focal or off‐target sites. Off‐target auditory effects, including transient hearing changes or tinnitus, have also been reported in some studies, likely due to cochlear stimulation via bone conduction [50]. These mechanisms should be systematically monitored through real‐time acoustic feedback and, where possible, passive cavitation detection.

5.5. Safety Considerations for Repeated and Home‐Based Exposure

The therapeutic vision for TUS—particularly the prospect of disease modification through repeated or chronic administration and the development of portable home‐use devices—raises important but largely unaddressed safety questions. Current evidence on TUS safety is derived almost exclusively from short‐term, single‐session, or limited‐session protocols. The safety profile of cumulative or repeated exposure over weeks, months, or years remains unknown. Critical gaps include: the potential for cumulative thermal or mechanical effects, the long‐term consequences of repeated modulation on neuronal and vascular integrity, and the absence of established monitoring protocols for home‐use settings. A recent consensus statement has emphasized that the safety of repeated TUS protocols and the use of portable home devices should be evaluated in dedicated long‐term safety studies before these applications become clinically available. Future trials should systematically collect and report long‐term adverse event data, employ standardized safety monitoring protocols, and investigate device modifications that would minimize the risk of damage—such as real‐time dosimetry, automatic shut‐off mechanisms, and patient‐specific acoustic modeling—to ensure patient safety as the field advances toward broader clinical translation.

6. Future Developments of TUS

6.1. Precision Neuromodulation Through Parameter Screening and Cell‐Type Targeting

TUS holds great promise for achieving precise neuromodulation in the brain—that is, selectively activating or inhibiting specific neural circuits based on their anatomical location and cell‐type composition. This precision can be achieved through two key advances: (1) high‐throughput parameter screening, which allows rapid testing of multiple ultrasound intensities, frequencies, and durations to identify optimal settings for each brain region; (2) cell type resolution, enabled by emerging technologies such as single‐cell RNA sequencing, optogenetics, and sonogenetics, which help map how different neuronal populations (such as excitatory and/or inhibitory neurons) respond to ultrasound. Together, these approaches will enable personalized TUS protocols that target specific brain circuits with minimal off‐target effects by leveraging its high spatiotemporal resolution and multimodal technologies, TUS is thus poised to play an important role in neuroscience research and in understanding neural circuits. The therapeutic spectrum will likely expand to include other neurodegenerative diseases, psychiatric and developmental disorders, as well as addiction and metabolic diseases.

6.2. Multimodal Integration: Pharmacomechanical Combinations, Co‐Stimulation, and Closed‐Loop Systems

Furthermore, TUS holds promise for integration with other treatment strategies: (1) Pharmacomechanical combinations: combining TUS with pharmacological agents such as drugs targeting neuroinflammation or synaptic plasticity may enhance its therapeutic effects through synergistic mechanisms; (2) integration with other neuromodulation techniques: integrating TUS with other neuromodulation techniques including tDCS and rTMS could enable multi‐target modulation of complex brain networks involved in conditions such as depression or PD; (3) closed‐loop system integration: real‐time feedback from fMRI or electroencephalography (EEG) could guide adaptive TUS, allowing dynamic adjustment of parameters based on ongoing brain activity, thus maximizing efficacy and minimizing side effects. This will enable intelligent parameterization, precise circuit targeting, diversified applications, and synergistic treatments with different modalities, ultimately reshaping the treatment paradigm of brain disorders. Importantly, this shift raises the possibility of moving beyond temporary symptom relief toward more durable effects. Preclinical evidence suggests that TUS can induce structural and functional reorganization of brain circuits; however, whether this translates into lasting recovery or disease modification in patients remains a key open question that requires rigorous long‐term clinical investigation.

6.3. Mechanotransduction‐To‐Oscillation Encoding: A Working Hypothesis and Forward‐Looking Roadmap

Beyond these technical pathways, we propose a working hypothesis—termed mechanotransduction‐to‐oscillation encoding—as a conceptual framework to help integrate existing findings and guide future investigation. It is important to emphasize that this framework is not yet empirically validated; rather, it is offered as a speculative organizing principle to generate testable predictions. The central idea is that the mechanical pressure waves delivered by TUS carry a precise temporal structure defined by the pulse repetition frequency (PRF), a key parameter determining the role of TUS. This temporal structure can be faithfully transduced by mechanosensitive ion channels into rhythmic fluctuations of membrane potential, thereby selectively shaping neural activity in specific frequency bands through resonance, entrainment, or disruptive interference with endogenous oscillations. In this view, the therapeutic effect of TUS depends not only on the spatial target and intensity, but more fundamentally on a nonlinear encoding relationship between PRF and the intrinsic oscillatory properties of the stimulated region. This perspective provides a unified interpretation for a range of otherwise fragmented parameter–effect observations: each effect reflects the predictable interaction between ultrasound frequency and local circuit resonance, rather than an isolated parameter–outcome association.

Distinguishing established evidence from speculation within this framework. Several lines of evidence support the plausibility of this framework. First, it is well established that TUS can modulate neural oscillations and that its effects depend on stimulation parameters such as PRF [10]. Second, mechanosensitive ion channels (e.g., Piezo1, TRPC6) have been demonstrated to transduce ultrasound mechanical signals into neuronal activity [6]. Third, concurrent electrophysiological recordings in humans have shown that different TUS protocols (e.g., tbTUS vs. 10 Hz TUS) can differentially modulate oscillatory activity in deep brain regions [4]. However, the specific claims of this framework—particularly, (1) the nonlinear encoding relationship between specific PRFs and intrinsic oscillatory frequencies across different brain regions, (2) the resonance or entrainment effects that would enable frequency‐specific modulation, and (3) the clinical utility of matching PRF to pathological rhythms—remain largely speculative and require direct experimental testing.

More importantly, the following research roadmap is therefore proposed not as a set of established clinical applications, but as a series of experimental tests designed to validate or falsify this hypothesis. These include: (1) systematically mapping PRF‐oscillation response functions across multiple brain regions using concurrent electrophysiological recordings to verify harmonic resonance and frequency entrainment; (2) testing oscillation‐frequency matching in disease models with characteristic pathological rhythms, such as PD (exaggerated pathological beta bursts) and AD (attenuated gamma rhythms), to establish disorder‐specific optimal parameters; (3) examining whether ultrasound pulse frequency is transmitted with high fidelity from the stimulation target to downstream brain regions, thereby evaluating the feasibility of mechanical frequency signals serving as information carriers in brain networks; (4) developing phase‐locked closed‐loop TUS systems that precisely lock ultrasound pulses to specific oscillation phases, enabling enhanced neuromodulatory effects without increasing energy delivery; (5) leveraging resting‐state oscillatory spectra and large‐scale TUS response datasets to train deep learning models that predict individually optimal PRF, thus enabling a precision stimulation strategy of read the oscillation first, then prescribe the frequency. The underlying logic of this strategy is to shift TUS parameter optimization from empirical trial and error screening toward rational frequency design grounded in the intrinsic oscillatory profiles of target circuits and disease‐specific pathological rhythms. This transition represents a critical step from spatially precise stimulation to dual spatial‐frequency precision neuromodulation and constitutes an indispensable encoding‐level dimension within the envisioned framework of hyper‐personalized closed‐loop multi‐target neuromodulation. While these prospects are conceptually promising, they remain speculative and require systematic validation through well‐powered, sham‐controlled, long‐term clinical trials before they can be considered clinically actionable.

7. Summary and Prospect

TUS has emerged as a transformative neuromodulation tool in the field of neurological and psychiatric disorders, primarily due to its non‐invasive nature, high spatial resolution, and the ability to selectively modulate deep brain regions. This modality uniquely allows for precise neural regulation between cortical and deep subcortical structures, offering unprecedented possibilities for interventions in complex brain neural networks. However, translating this potential into widespread and reliable clinical applications faces several critical challenges. Currently, the efficacy of TUS is influenced by various factors, including individual skull anatomical variations, stimulation parameters, and physiological states of the target areas. The underlying mechanisms remain incompletely understood, with key aspects including the activation of mechanosensitive ion channels, thermal effects, and the determination of safe cavitation thresholds under low‐intensity conditions. These uncertainties result in a lack of standardized treatment protocols and necessitate the improvement of individualization and precision.

Notably, neuroimaging technologies play an indispensable role in the clinical translation of TUS therapy. By integrating imaging modalities such as MRI and PET, researchers can non‐invasively monitor changes in brain function and structure before and after TUS treatment. This allows for the elucidation of therapeutic mechanisms, the identification of potential biomarkers to select appropriate treatment populations, and the dynamic tracking of treatment responses. This imaging‐therapy integrated research paradigm not only aids in a deeper understanding of the regulatory mechanisms of TUS on brain disorders but also provides critical support for the realization of individualized and precision TUS treatments. In the future, with the deep integration of multimodal imaging technologies and TUS, it is anticipated that TUS will move from experimental research to standardized clinical applications, truly achieving the clinical translation goal of image‐guided therapy.

In conclusion, future developments should be guided by multidisciplinary integration, combining research findings from acoustics, neuroscience, imaging, and clinical medicine. The primary task is to establish evidence‐based standardized stimulation protocols and utilize emerging technologies such as artificial intelligence to dynamically optimize parameters. Additionally, it is essential to delve into the molecular and circuit mechanisms of TUS through fundamental research and systematically evaluate its efficacy and safety through long‐term, multicenter clinical trials. Only through these efforts can TUS transition from a promising technology to a robust therapeutic modality, fundamentally altering the treatment landscape for neurological and psychiatric disorders.

Author Contributions

Yunjie Sui: writing – original draft, conceptualization, investigation. Mingyue Wei: data curation, writing – review and editing. Robert Chen: writing – review and editing. Xue Xia: resources, project administration, writing – review and editing.

Funding

This work was supported by grants from the National Natural Science Foundation of China (no. 32300921), Special Funding for Taishan Scholars Project (no. tsqn202312244), Key R&D Program of Shandong province (no. 2025RZB0804), and the Natural Science Foundation of Shandong Province (no. ZR2026MS0429).

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.