Feasibility and Safety of Transcranial Focused Ultrasound on Language Impairment in a Patient with Autism Spectrum Disorder: A Case Report

Abstract

Background

Autism spectrum disorder (ASD) is defined by persistent difficulties with social communication and language, and a subset of children stay minimally verbal despite intensive interventions. The left amygdala and ventral-anterior superior temporal sulcus (STSva) form a key circuit for social-language integration, but how transcranial focused ultrasound (tFUS) might modulate in minimally verbal ASD has not been explored.

Case Presentation

An 11-year-old minimally verbal ASD boy received 4 weeks of sequential tFUS stimulation (0.5 MHz). The treatment targeted the left amygdala and STSva, five times per week. Outcome assessments included the Mandarin Clinical Evaluation of Language for Preschoolers’ Core Scale (MCELP-CS), Autism Behavior Checklist (ABC), and functional near-infrared spectroscopy (fNIRS) at pre-tFUS, post-tFUS, and a 4-week follow-up. The boy tolerated the intervention well, with no adverse events. Significant improvements in key language skills (vocabulary naming +10%, sentence structure imitation +22%) and reduced ASD symptoms (ABC language −40%, social interaction −26%) were observed post-tFUS and sustained at follow-up. fNIRS confirmed altered cortical activation and reorganized functional connectivity in language-related brain regions.

Conclusion

This single-case observational report describes acceptable tolerance to sequential tFUS targeting the left amygdala-STSva circuit in a minimally verbal child with ASD, alongside concomitant behavioural and neuroimaging changes. These preliminary observations provide descriptive, hypothesis-generating data for this circuit-directed intervention strategy and highlight the necessity of larger controlled trials to further evaluate its potential clinical utility.

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental condition marked by difficulties in social communication and repetitive behaviors,¹ with language impairment as a core feature.² It is important to distinguish isolated linguistic deficits from pervasive social‑communication impairment: a substantial subset of ASD children with minimally‑verbal status struggle not only with linguistic form, but also with socially‑driven communicative engagement.³ Notably, around 25–30% of children with ASD remain minimally verbal even after intensive interventions.⁴ This clinical presentation cannot be fully explained by damage to classical language‑specialized cortical regions alone; a core deficit lies in atypical processing of socially cues, which relies on a neural network including the left amygdala⁵ and left ventral-anterior superior temporal sulcus (STSva).⁶ Functional magnetic resonance imaging (fMRI) meta-analyses reveal hypoactivation in the left superior temporal sulcus (STS) during perception of biological motion and speech-related stimuli, as well as altered amygdala responsiveness to social stimuli,⁷ and weakened functional connectivity (FC) between the amygdala and STS in ASD—deficits that correlate with social communication domains rather than purely linguistic performance.⁸

Transcranial focused ultrasound (tFUS) is a non-invasive neuromodulation technique with high spatial resolution for both superficial and deep neural structures,⁹ capable of transiently altering neuronal excitability via mechanosensitive ion channels (at intensities below safety thresholds, mechanical index <1.9) to promote network plasticity.¹⁰ These mechanistic properties of tFUS are well‑documented in pre‑clinical models. Nevertheless, clinical evidence supporting tFUS application in ASD remains limited and preliminary, with no established optimal target configuration or standardized protocol for minimally verbal populations. However, to date, human tFUS investigations among participants with ASD have predominantly targeted superficial cortical regions such as the left dorsolateral prefrontal cortex.¹¹ No prior human studies have implemented sequential stimulation protocols aiming to modulate the deep left amygdala together with STSva. Against this background, this case study investigates the feasibility and tolerability of sequential tFUS stimulation of the left amygdala and STSva in a minimally verbal ASD participant.

Case Presentation

This participant is an 11-year-old male diagnosed with ASD based on the diagnostic criteria in the Diagnostic and Statistical Manual of Mental Disorders (5^th Edition), with this diagnosis confirmed by a pediatric medical center in Shanghai. The patient has no family history of ASD, neurodevelopmental disorders, or other neurological diseases in first-degree relatives (parents and siblings), and no relevant genetic abnormalities were detected via routine genetic screening for ASD. The participant has severe language comprehension delay (language comprehension ability development age is 5–6 years old) and extremely severe language expression function delay (language expression ability development age is 4 years old), despite having received 3 years of conventional behavioral and speech-language intervention (one-on-one speech training, applied behavior analysis) at a professional rehabilitation institution. The prior conventional interventions were administered 5 times per week, 40 minutes per session, with no meaningful improvement in language comprehension or expressive abilities—core language deficits persisted, which was the primary reason for seeking exploratory neuromodulation treatment.

The participant has no metal foreign bodies or other implanted electronic devices in the body, has not received other neuroregulatory treatments in the past 3 months, and no history of epilepsy, head trauma, or other neurological diseases. Neurological examination did not reveal any focal defects; visual and auditory acuity tests were normal, and no motor disorders were identified—all factors that would not hinder cooperation during examination and treatment.

The patient is right-handed. At the baseline assessment, the participant exhibited expressive language delay, limited practical communication skills, and reduced response ability to social cues. Given that language disorders persisted in the absence of structural neurological abnormalities, and the patient does not have severe visual, auditory, deaf-mutism, or motor disorders (which would hinder cooperation during the examination and treatment), the participant is considered suitable for exploratory neuroregulatory treatment targeting the neural networks related to language. Based on previous evidence indicating that the superior temporal sulcus plays an important role in social perception and language integration,¹¹ the amygdala manages the processing of information between the prefrontal-temporal association cortices and the hypothalamus. Therefore, the tFUS technology targeting the left amygdala and STSva was selected.

This study was registered on the Chinese Clinical Trial Registry website (ChiCTR) (website: www.chictr.org.cn; registration number: ChiCTR2500096862), and it followed the provisions of the Helsinki Declaration. A written informed consent form was obtained from the legal guardian of the participant before the start of the study, and a separate written informed consent for the publication of this case report was also obtained from the mother. The research protocol was approved by the Medical Ethics Committee of Shanghai YangZhi Rehabilitation Hospital (Shanghai Sunshine Rehabilitation Center) (approval number [2024] 032).

Prior to intervention, a formal risk‑benefit assessment was completed by the multidisciplinary research team. Given the experimental status of dual‑target tFUS for paediatric ASD, potential risks including scalp heat, headache and so on were comprehensively discussed with the child’s legal guardian. The anticipated exploratory benefit was to collect preliminary observational data on the tolerability of this circuit‑targeted protocol, rather than guaranteeing clinical improvement. The guardian gave voluntary written informed consent after full disclosure of experimental uncertainties.

tFUS Intervention Protocol

tFUS was delivered using the Sonic Concepts NeuroFUS PRO™ system (0.5 MHz) with neuro-navigation system (NDI Polaris Lyra®). Throughout the entire intervention period, strict real-time safety monitoring was implemented. Real‑time behavioural observation was performed during each tFUS session. After each tFUS treatment, a 10-minute on-site observation was required to record any immediate adverse reactions, including scalp discomfort, headache, dizziness, irritability, sleep disorders or abnormal behavioral outbursts. The mother was also asked to record the child’s emotional state, sleep, diet and social behavior every night during the 4-week intervention period and the subsequent 4-week follow-up period. This study utilized the BabelBrain v0.8.000 software to incorporate T1-weighted magnetic resonance imaging (MRI) scans into the neuro-navigation system, and completed transcranial ultrasound and thermal model analysis, as shown in Figure 1A. Stimulation parameters all followed the safety standards for head ultrasound set by the US Food and Drug Administration (FDA) (Isppa<190 W/cm², Ispta<720 mW/cm²).¹²

• First, stimulate left amygdala (inhibitory): Isppa = 30 W/cm², Ispta = 0.48 W/cm², pulse repetition frequency (PRF) = 40 Hz, duty cycle (DC) = 1.6%, pulse train duration = 400 ms, single pulse duration = 0.4 ms, pulse repetition interval = 25 ms, 30 min/session

• Then stimulate the left STSva (excitatory): Isppa = 12 W/cm², Ispta = 0.72 W/cm², PRF = 200 Hz, DC = 6%, pulse train duration = 1000 ms, single pulse duration = 0.3 ms, pulse repetition interval = 5 ms, 20 min/session

Figure 1.

tFUS neuromodulation planning, experimental setup, and resting-state fNIRS channel layout and functional connectivity alterations across pre- and post-tFUS timepoints. (A) The babelbrain software ensured that all tFUS parameters strictly adhere to the FDA’s guidelines for cranial ultrasound safety. (B) Experimental setup for individualized tFUS targeting guided by structural imaging. (C) Distribution of 24 fNIRS channels and their corresponding regions of interest (ROIs). (D) Only FC between ROIs with a Pearson correlation coefficient > 0.5 are visualized.

Abbreviations: AAL, Automated Anatomical Labeling atlas; FC, functional connectivity; L, Left; R, Right; PFC, Prefrontal Cortex; SMC, sensorimotor cortex; IFG, inferior frontal gyrus; ROI, region of interest.

The treatment is performed daily (Monday to Friday) and lasts for 4 weeks (a total of 20 treatment sessions) with 100% adherence. Tolerability was assessed via real-time observation during each session (for signs of discomfort, restlessness, or adverse reactions) and 10-minute post-session monitoring; no adverse reactions (such as headache, skin irritation, behavioral deterioration, etc) were reported.

T1-weighted MRI is used to determine the target brain regions of the subjects and to guide the transducer to the required positions in the neuro-navigation system,¹³ as shown in Figure 1B. Written informed consent has been obtained from the child’s mother, allowing the photos to be published in Figure 1B.

Outcome Assessments

• Mandarin Clinical Evaluation of Language for Preschoolers’ Core Scale (MCELP-CS): Evaluatedvocabulary comprehension (VC), vocabulary naming (VN), sentence comprehension (SC), sentence structure imitation (SSI), and story narration (SN).¹⁴ In addition, the original psychometric validation study of M-CELP-CS included a one-month retest interval and demonstrated acceptable measurement stability (intra-class correlation coefficients = 0.82).

• Autism Behavior Checklist (ABC): Assessed ASD symptoms across five subscales: sensory, relationship building, body and object use, language, and self-care.¹⁵

• Functional near-infrared spectroscopy (fNIRS): Hitachi ETG-4100 system (10 Hz sampling rate, 695/830 nm wavelengths) measured resting-state FC in 24 prefrontal channels. According to the Automated Anatomical Labeling (AAL) partition, 24 channels were divided into 6 regions of interest (ROIs), as shown in Figure 1C. Data were preprocessed using NIRS_KIT.¹⁶ During the data collection process, the participant remained quiet and in a fixed position. In the data processing stage, in order to eliminate high-amplitude spikes and baseline drifts caused by muscle movements in the head or face, we adopted the Temporal Derivative Distribution Repair (TDDR) method. One of the significant advantages of this method is that it can effectively remove both spike and baseline drift motion artifacts simultaneously without requiring users to set any threshold parameters.¹⁷

Assessments were performed at pre-tFUS, post-tFUS, and 4-week follow-up. Behavioral scales were scored by two independent blinded raters.

Treatment Course and Outcome

During the 4-week intervention period, the mother and the rehabilitation therapist accompanied the child throughout the process. The participant continued his regular behavioral and speech-language rehabilitation interventions at unchanged frequency and intensity throughout the 4-week tFUS intervention and the subsequent 4-week follow-up period. The child cooperated well during each treatment session, and all 20 treatments were completed 100%. After each treatment, we conducted a 10-minute real-time monitoring, and no headache, skin irritation, behavioral changes, or mental symptoms were observed; the participants maintained a stable daily routine and no new emotional fluctuations, stereotyped behaviors, or mental symptoms occurred.

Table 1 summarizes the changes in the scores of the core scale for assessing the language abilities of the MCELP-CS and the ABC before and after tFUS intervention. MCELP-CS scores showed improvements in: VN (+10% post, +18% follow-up), SSI (+22% post, +18% follow-up), and SN (+45% post, +50% follow-up). VC and SC showed minimal or delayed improvements, consistent with the intervention’s focus on expressive language and social-linguistic integration. ABC total scores decreased from 62 (pre-tFUS) to 46 (post-tFUS, −26%) and 39 (follow-up, −37%). Key subscale changes included relationship building (R), language (L), sensory (S1).

Table 1.

Changes in Language and Behavioral Outcomes Before and After tFUS

Measure	Pre-tFUS	Post-tFUS	4-Week Follow-up	Acquisition Change (%)
				(Post – Pre)/Pre	(Follow-up–Pre)/Pre
MCELP-CS
VC	82.86%	85.71%	82.86%	+3	0
VN	60.00%	66.15%	70.77%	+10	+18
SC	78.26%	78.26%	91.3%	0	+17
SSI	50%	60.83%	59.17%	+22	+18
SN	40%	58%	60%	+45	+50
ABC	62	46	39	−26	−37
S1	3	1	1	−67	−67
R	23	17	11	−26	−52
B	6	6	5	0	−17
L	17	11	11	−35	−35
S2	13	11	11	−15	−15

Abbreviations: tFUS, transcranial focused ultrasound; MCELP-CS, Mandarin Clinical Evaluation of Language for Preschoolers’ Core Scale; VC, Vocabulary Comprehension; VN, Vocabulary Naming; SC, Sentence Comprehension; SSI, Sentence Structure Imitation; SN, Story Narration; ABC, Autism Behavior Checklist; S1, Sensory; R, Relationship building; B, Body and Object Use; L, Language; S2, Self-care.

fNIRS revealed reorganized FC post-tFUS (see Table 2). Several connections showed increased functional coupling after treatment: LPFC–LSMC, RPFC–RPFC, and RIFG–RPFC. In contrast, connectivity decreased in several fronto-parietal and fronto-temporal pathways: LIFG–LPFC, LIFG–RIFG, LSMC–RPFC, LPFC–RPFC, LPFC–RIFG, and RIFG–RSMC. Heatmaps, see Figure 1D, revealed a shift toward more balanced network dynamics, with enhanced bilateral frontal-motor areas and reduced hyperconnectivity in certain lateral temporal and parietal networks—consistent with improved language and social behavior.

Table 2.

Analysis of FC

Channel-Channel	ROI-ROI	FC Difference (Post-tFUS – Pre-tFUS)
CH10-CH11	LPFC-LSMC	0.5
CH13-CH15	RPFC-RPFC	0.58
CH14-CH15	RIFG-RPFC	0.65
CH3-CH9	LIFG-LPFC	−0.6
CH3-CH19	LIFG-RIFG	−0.5
CH8-CH16	LSMC-RPFC	−0.52
CH9-CH16	LPFC-RPFC	−0.57
CH9-CH19	LPFC-RIFG	−0.55
CH17-CH22	RIFG-RSMC	−0.64

Abbreviations: CH, fNIRS channel; FC, functional connectivity; tFUS, transcranial focused ultrasound; ROI, region of interest; L, Left; R, Right; PFC, Prefrontal Cortex; SMC, sensorimotor cortex; IFG, inferior frontal gyrus.

Discussion

This single-case study evaluated whether sequential tFUS targeting the left amygdala and STSva improves outcomes in a minimally verbal ASD child. Core findings— language gains, reduced ASD symptoms, and altered cortical connectivity—supports the clinical feasibility of targeting a social–language integration circuit rather than a single cortical language node.

The strongest language gains were seen in VN, SSI, and SN, with several effects sustained or further improved at follow-up. This profile is compatible with the study’s network rationale that communicative learning in ASD depends not only on “core language” representations but also on the extraction of social salience and communicative intent—processes linked to amygdala–STS circuitry.⁶ Existing studies have reported heterogeneous connectivity characteristics in ASD, which are affected by factors such as disease severity.¹⁸ A plausible interpretation is that during the treatment, by first inhibiting a deep affective salience hub (amygdala), and then activating a multimodal integration hub (STS/STSva), the hyperexcitability of the amygdala¹⁹ can be regulated, and the hypofunctional STSva²⁰ can be activated. This produces a synergistic regulatory effect on the entire circuit, possibly enhancing the signal value and integration of social-related language input, thereby prioritizing the development of expressive language and imitation—skills that are often particularly limited in minimally verbal children with ASD.²¹ STS is regarded as a central hub for social perception and cognition,²² playing a role in processing facial cues and human movements.⁶ Kausel et al highlighted the left STS’s potential relevance in understanding neurodevelopmental conditions with social communication challenges, including ASD.²³ For this study, the MCELP-CS was chosen specifically to reduce floor effects in a school-aged child with preschool-level language. This choice enhances the ability to detect changes²⁴ but also limits cross-age normative interpretation.¹⁴

Reductions in ABC scores, especially in the relationship building and language subscales, point to meaningful improvements in how the child functions in daily social and communicative contexts. These objective changes were corroborated by the mother’s report of increased spontaneous speech and social responsiveness, confirming real-world relevance.

For neuroimaging selection, this study relied on fNIRS to record prefrontal cortical signals, even though the primary stimulation targets lay in deeper subcortical and temporal regions. Importantly, fNIRS only measures hemodynamic signals from superficial cortical layers and cannot directly capture neural activity within the deep stimulation targets. The prefrontal cortex is connected with the amygdala²⁵ and the amygdala is connected with the STS.²⁶ This indicates that changes in the connectivity of the prefrontal cortex can indirectly reflect the network-level plasticity triggered by deep tFUS neural regulation. fNIRS showed a redistribution of connectivity rather than uniform increases, consistent with adaptive network plasticity in ASD.^7,8 The dual-target strategy represents a novel approach compared with prior tFUS studies targeting only cortical regions.¹¹ The present approach is more directly grounded in ASD pathophysiology implicating amygdala and STS abnormalities, and STS serves as a crucial “hub” in the process of social information processing. Alongside this, mechanistic research has demonstrated that tFUS is capable of regulating neuronal excitability²⁷ and cortical plasticity²⁸ through mechanosensitive pathways. The current study’s explicit reporting of intensity and duty cycle enhances the interpretability and addresses a common limitation in early neuromodulation case reports.

Limitations include the single-case pre–post design without a sham control, fNIRS depth constraints, and dual-target mechanistic ambiguity. These highlight the need for controlled factorial trials to validate efficacy. We also acknowledge the limitations brought about by the assessment time points. This participant, after 3 years of conventional behavioral and speech-language intervention, still had severe and persistent language disorders with almost no improvement; therefore, the rapid functional improvement observed after tFUS is unlikely to be solely due to the repetitive neuropsychological tests. In future studies, we will extend the follow-up period to track the persistence of the treatment effect. Furthermore, the overall evidence base for sequential dual‑target tFUS neuromodulation applied to the left amygdala‑STSva circuit among minimally verbal children with ASD remains limited. Therefore, all findings reported herein should be interpreted strictly as preliminary observational data.

Conclusion

Neuromodulation of the left amygdala‑STSva circuit using tFUS was feasible and well‑tolerated in this single case of a minimally verbal child with ASD. Descriptively, this circuit-targeted intervention may be associated with preliminary improvements in language and social‑communication skills. Given the uncontrolled single-case study design, no causal inference or definitive therapeutic conclusion can be drawn. These tentative observational findings lend preliminary, hypothesis-generating support to this circuit‑oriented intervention strategy and highlight the requirement for larger controlled trials to rigorously verify its clinical efficacy.

Funding Statement

This research was funded by the National Key Clinical Specialty Discipline Construction Project of China (Z155080000004), the Shanghai Research Center of Rehabilitation Medicine (Top Priority Research Center of Shanghai) (2023ZZ02027), the Shanghai Municipal Health and Wellness Commission Health Industry Clinical Research Specialization (202340119), the China Disabled Persons’ Federation (2024CDPFHS-16), and the Songjiang Key Science and Technology Research Project (2024SJKJGG120).

Abbreviations

ABC, Autism Behavior Checklist; CH, fNIRS channel; FC, functional connectivity; IFG, inferior frontal gyrus; L, Left; MCELP‑CS, Mandarin Clinical Evaluation of Language for Preschoolers’ Core Scale; PFC, Prefrontal Cortex; R, Right; ROI, region of interest; S1, Sensory; S2, Self‑care; SC, Sentence Comprehension; SMC, sensorimotor cortex; SSI, Sentence Structure Imitation; SN, Story Narration; tFUS, transcranial focused ultrasound; VC, Vocabulary Comprehension; VN, Vocabulary Naming.

Data Sharing Statement

The datasets used and/or analyzed during the current study are available from Beihua Zhang on reasonable request.

Ethics Approval and Consent to Participate

The study was carried out in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee of Shanghai Yangzhi Rehabilitation Hospital (Shanghai Sunshine Rehabilitation Center) (approval number: [2024] 032). According to the hospital’s policy, publishing a single anonymous case report did not require additional approval from the hospital. Informed consent was obtained from the mother. Written informed consent for the publication of this case report was also provided by the mother.

Consent for Publication

Informed consent was obtained from the mother of the patient participating in the study. Additionally, the publication of this paper also received written informed consent from the mother of the patient.

Disclosure

The authors declare no conflicts of interest.