Right globus pallidus ultrasound shows a delusion signal in schizophrenia
A small sham-controlled crossover trial points to the right globus pallidus interna as a plausible neuromodulation target in schizophrenia. In ten outpatients with persistent psychosis, MRI-guided transcranial ultrasound stimulation produced a greater reduction in delusional symptoms than sham, while auditory hallucination scores and cognitive measures did not separate clearly between conditions.
The finding is modest in size, but clinically meaningful in direction. Delusions, more than hallucinations in some patients, often reflect a network-level failure of salience assignment and cognitive control rather than a single focal lesion, so a circuit intervention aimed at the basal ganglia has a biologically coherent rationale. Here, investigators targeted a deep subcortical node that sits inside the cortico-striatal-pallido-thalamic loop, a pathway repeatedly linked to psychosis severity and treatment response.
A deep-brain target reached without surgery
The trial used a randomized, sham-controlled, crossover design with one active session and one sham session separated by a week. Ten outpatients with schizophrenia or schizoaffective disorder participated; the mean age was 37.5 years, half were female, and baseline psychosis burden was substantial, with a mean BPRS total score of 64.4 and a positive-symptom subscale score of 21.8. Six participants were taking antipsychotic medication and remained on the same dose throughout the study, while four were not receiving antipsychotic therapy during the trial.
A 350 kHz transducer delivered pulsed stimulation for 10 minutes, using a 100 Hz pulse repetition frequency, a 5% duty cycle, and MRI-based neuronavigation to focus on the right GPi. The protocol was engineered with a deliberate emphasis on dose control and targeting precision, and the authors first validated the acoustic field in free water and then measured skull transmission through human skull specimens. The reported pressure attenuation through the skull was substantial, yet the estimated in situ intensity remained within the planned low-intensity range.
One session is not a treatment course.
That matters because this design tests biologic plausibility, not clinical durability. Even so, it provides an unusually direct look at whether a noninvasive energy-based intervention can reach a deep motor-limbic node and alter a psychiatric symptom domain that has resisted many pharmacologic and behavioral strategies.
Delusion scores improved more after active stimulation than sham
The prespecified symptom measure for delusional experience changed in favor of active stimulation. Total DESA scores decreased after transcranial ultrasound, with a mean change of −1.2 after TUS and 0.4 after sham. In the mixed-effects model that incorporated all post-baseline observations and adjusted for treatment, time, and session order, TUS was associated with a significantly greater reduction in total DESA score than sham, with an estimate of −1.56, a 95% confidence interval from −2.54 to −0.58, a p value of 0.002, and Cohen’s d of 0.68.
The auditory hallucination measure did not show a significant treatment effect. Neither did the cognitive assessment, although the Montreal Cognitive Assessment trended only narrowly away from significance. In practical terms, the signal was selective rather than broad, which is often what one sees when a circuit intervention hits a network node more closely tied to a specific symptom dimension than to the full psychosis phenotype.
The time course is especially interesting. Exploratory modeling suggested the largest estimated difference between TUS and sham appeared at day 7, not immediately after stimulation. That delayed peak fits with a neuromodulatory mechanism that extends beyond simple on-off inhibition. In the discussion, the investigators connect this pattern to the possibility of offline plasticity, including changes in glutamatergic signaling and calcium-dependent receptor behavior. For schizophrenia, where maladaptive circuit synchronization can persist despite dopamine blockade, a delayed network recalibration is a more compelling hypothesis than a transient sensory effect.
Safety, sham blinding, and the limits of a ten-patient pilot
No serious adverse event occurred. Mild transient effects were reported in both groups, including fatigue, headache, gastrointestinal discomfort, and in some cases increased delusions or hallucinations on side-effect surveillance. Seven participants after TUS and all ten after sham endorsed at least one item on the systematic adverse-effect checklist, but the pattern did not suggest a dangerous safety signal.
Blinding was imperfect but not obviously broken. Six participants correctly identified the stimulation condition after the study, a rate not distinguishable from chance. An exploratory sensitivity analysis did not find that correct guessing explained the DESA change, nor did it show evidence that the treatment effect differed by guess status. The crossover sequence also did not show a clear carryover pattern, although the sample was far too small for strong inference on that question.
Several limitations remain hard to avoid. The study enrolled only ten people, so confidence intervals are wide and the day-7 effect, while intriguing, does not survive the sort of correction that larger trials require. Baseline symptom burden was high, but the cohort was heterogeneous with respect to antipsychotic exposure, and the intervention was tested in a single-session format. The most defensible reading is not that GPi-targeted ultrasound has already become a treatment, but that it has crossed an important threshold: it appears feasible, tolerable, and capable of altering delusional severity in a direction worth following.
Future work will need a larger randomized sample, more rigorous symptom phenotyping, and longer follow-up to determine whether the apparent effect persists, broadens, or disappears. The key unresolved question is whether right GPi stimulation can be developed into a reproducible circuit-based intervention for a subset of patients whose psychosis is driven by a particular pattern of basal ganglia dysfunction rather than by schizophrenia in general.