Transcranial focused ultrasound has quietly become one of the most tantalizing tools in modern neuroscience. By concentrating sound waves through the skull and into precisely defined brain regions, the technique promises to treat tumors, modulate neuronal activity, open the blood-brain barrier for drug delivery, and potentially ease the symptoms of Alzheimer’s disease, Parkinson’s disease, and depression, all without a single incision. Yet every one of those applications depends on a deceptively difficult question: what actually happens to the ultrasound beam on its journey through the human head? Because directly measuring pressure fields inside a living brain is impossible, researchers rely on computer simulations to predict where the acoustic energy will land. A new study published in Heliyon now shows that a widespread simplification in those simulations, treating the brain’s soft tissues as if they were little more than water, can measurably distort the predicted ultrasound focus.
The research team, led by Ji-Hun Yu, Woong-chang Yoon, Eun-Hee Lee, and Hyeon Seo, working with data from the Alzheimer’s Disease Neuroimaging Initiative, set out to quantify exactly how much modeling error is introduced when non-skull tissues are neglected. Their approach was systematic and unusually comprehensive. Drawing on T1-weighted and T2-weighted magnetic resonance images from twenty cognitively normal adults with a mean age of 65.4 years, the team segmented each head into nine distinct tissue types using the CHARM algorithm from SimNIBS: white matter, gray matter, cerebrospinal fluid, scalp, eyes, compact bone, spongy bone, blood, and muscle. Each tissue was assigned acoustic properties, including speed of sound, density, and attenuation coefficient, drawn from the IT’IS Foundation database, a standard reference in the field.
From these segmentations the researchers built four competing head model configurations of increasing anatomical fidelity. The simplest, designated S, contained only the skull immersed in water, a common approach that treats everything outside and inside the cranium as a homogeneous coupling medium. The second added the scalp, the third added the grouped non-skull soft tissues such as gray matter, white matter, cerebrospinal fluid, blood, eyes, and muscle, and the fourth, the most realistic, included both scalp and soft tissues. Simulations were then run with the open-source k-Wave toolbox, which solves the coupled first-order acoustic wave equations using a k-space pseudo-spectral time-domain method. To keep numerical error in check, the team resampled the MRI data from one-millimeter to 0.6-millimeter resolution, corresponding to roughly ten grid points per wavelength, well above the six points per wavelength generally considered sufficient for acoustic neuromodulation work.
The scale of the computational campaign is one of its distinguishing strengths. A custom single-element focused transducer operating at 250 kilohertz, with a radius of curvature of 110 millimeters and an aperture diameter of 128 millimeters, was positioned at thirteen target sites defined by the international 10-10 electroencephalography system, spanning seven coronal and seven sagittal locations. Each transducer was placed normal to the local skull surface, sixty millimeters away, to minimize aberration from oblique incidence. Multiplying twenty subjects by thirteen positions by four model configurations yielded a total of 1,040 full three-dimensional wave propagation simulations, plus an additional free-water run for each case to normalize the results. Few previous studies have attempted to isolate tissue-layer effects across such a broad matrix of anatomies and stimulation sites.
The headline finding is that the soft tissues matter more than many modelers have assumed. Normalized peak pressures across all conditions ranged from roughly 0.4 to 0.8, meaning the skull and overlying tissues attenuated the beam by 20 to 60 percent relative to free-water propagation. When the researchers compared models with and without the intracranial soft tissues, statistically significant differences in peak pressure appeared at eleven of the thirteen stimulation positions, whereas adding the scalp alone produced significant differences at only six to eight positions. In other words, the brain’s internal anatomy, not just the skull, exerts a measurable influence on how much acoustic pressure reaches the target. The effect was especially pronounced along the sagittal plane, where the propagation path crosses a denser thicket of tissue interfaces, producing reflections and refractions that can either reduce or, counterintuitively, slightly increase the focal pressure depending on the individual and the site.
Focal volume, defined as the region enclosed by the full width at half maximum of the pressure distribution, told a complementary story. Across the stimulation positions it ranged from about 500 to 3,000 cubic millimeters, and models that incorporated the soft tissues generally produced larger, more diffuse focal regions than their skull-only counterparts. This inverse relationship with peak pressure follows directly from the physics: a beam that loses pressure to attenuation and impedance mismatches also spreads its remaining energy over a broader zone. The spatial overlap between focal regions from different model configurations, quantified with an intersection-over-union metric across 1,560 pairwise comparisons, reinforced the pattern. Comparisons differing only in scalp inclusion produced tightly clustered overlap values between 0.70 and 0.98, while comparisons involving the soft tissues dropped to roughly 0.50 to 0.90, indicating that the internal anatomy can shift the very location and shape of the ultrasound focus.
The study also revisited an old question about which anatomical measurements best predict ultrasound delivery. Using the scalp-and-skull model, the team correlated tissue thicknesses, measured within a cylindrical volume aligned with the transducer axis, against simulation outcomes across 260 subject-position pairs. Skull thickness emerged as the strongest predictor, showing a significant negative correlation with peak pressure (r = -0.5334) and a significant positive correlation with focal volume (r = 0.5106), both at p < 0.01. Scalp thickness, by contrast, was only weakly associated with either outcome. Yet when the same correlation analysis was repeated within individual subjects across the thirteen positions, the relationships became statistically inconsistent, appearing clearly in only nine of twenty subjects for peak pressure. The authors attribute this to the crudeness of using a single average thickness value, which cannot capture skull curvature, surface roughness, or the internal heterogeneity of cortical and cancellous bone, and to the limited number of positions per subject.
Why does all of this matter beyond the modeling community? The clinical stakes are considerable. Low-intensity transcranial focused ultrasound is being explored as a noninvasive alternative to deep brain stimulation, avoiding the surgical implants that complicate patient compliance, and it offers millimeter-scale targeting that electromagnetic techniques like transcranial magnetic stimulation cannot match. Dosage in these applications is prescribed in terms of pressure and intensity at the target, so a simulation that overestimates focal pressure, or misplaces the focus by several millimeters, could translate directly into underdosed or misdirected treatment. Earlier work had already hinted at the problem: a study by Slominski and colleagues found that including cerebrospinal fluid in head models could raise the focal absorbed power density by up to 29 percent, and Guo and colleagues estimated that the brain itself contributes roughly 34 percent of total acoustic attenuation, far from negligible compared with the skull’s 83 percent.
The authors are careful to note the limitations of their own approach. All simulations were conducted at 250 kilohertz, a frequency chosen to make 1,040 runs computationally feasible and one that is widely used in clinical and artificial-intelligence-driven tFUS studies. But aberration and attenuation both grow with frequency, so the tissue-layer effects documented here are likely to become more pronounced, not less, at the higher frequencies used in some therapeutic applications. The cohort was also relatively old, with a mean age of 65.4 years, and skull morphology changes with age, so the findings may not fully generalize to younger populations. Fixed literature-based acoustic properties, moreover, cannot capture inter-subject variability, and the sensitivity analysis is sobering: to keep peak pressure error within 5 percent, the speed of sound must be accurate to within 1.75 percent for the skull and 2.5 percent for the scalp.
The practical takeaway is a call for richer, more individualized head models. Because acquiring CT scans purely for research participants raises radiation concerns, many laboratories have turned to MRI-based modeling, and this study suggests that such models gain real fidelity when they explicitly segment and assign properties to the soft tissues rather than collapsing them into water. The ideal future, the authors argue, lies in co-registered CT-MRI datasets that combine CT’s detailed view of skull heterogeneity with MRI’s rendering of intracranial anatomy, supported by experimental validation using multilayered phantoms that mimic the scalp, skull, and brain. Until then, the message for anyone planning a transcranial focused ultrasound experiment or trial is clear: the skull may be the great gatekeeper of cranial acoustics, but the brain behind it is far from acoustically invisible, and treating it as such risks steering the beam away from its intended target.
Subject of Research: Quantitative assessment of how scalp and intracranial soft tissue layers affect transcranial focused ultrasound simulation accuracy
Article Title: Quantitative impact of tissue layer composition in transcranial focused ultrasound modeling
Article References: Yu, J.-H., Yoon, W.-C., Lee, E.-H., & Seo, H. (2026). Quantitative impact of tissue layer composition in transcranial focused ultrasound modeling. Heliyon, 12(15), Article e45487. https://doi.org/10.1016/j.heliyon.2026.e45487
Image Credits: AI Generated
DOI: 10.1016/j.heliyon.2026.e45487
Keywords: transcranial focused ultrasound, neuromodulation, acoustic simulation, k-Wave, skull modeling, MRI segmentation, brain tissue, peak pressure, focal volume, blood-brain barrier, Alzheimer’s disease, head modeling
Cite Scienmag News
APA MLA Chicago
Cassandra Pierce. (September 30, 2026). Brain Tissue Layers Reshape Ultrasound Beams in Transcranial Stimulation Models. Scienmag. https://scienmag.com/brain-tissue-layers-reshape-ultrasound-beams-in-transcranial-stimulation-models/
Cassandra Pierce. “Brain Tissue Layers Reshape Ultrasound Beams in Transcranial Stimulation Models.” Scienmag, 30 September 2026, https://scienmag.com/brain-tissue-layers-reshape-ultrasound-beams-in-transcranial-stimulation-models/. Accessed 30 September 2026.
Cassandra Pierce. “Brain Tissue Layers Reshape Ultrasound Beams in Transcranial Stimulation Models.” Scienmag. September 30, 2026. https://scienmag.com/brain-tissue-layers-reshape-ultrasound-beams-in-transcranial-stimulation-models/
Copy citation Download RIS
Tags: acoustic simulationAlzheimer’s diseaseblood-brain barrierblood-brain barrier opening with ultrasoundbrain tissuebrain tissue layer effects on ultrasound beam focusingbrain tissue properties in ultrasound modelingcomputer simulations of transcranial focused ultrasoundeffects of brain tissue layers on ultrasound beam accuracyfocal volumehead modelingimpact of brain tissue heterogeneity on ultrasound targetingk-Wavemodeling errors in ultrasound brain therapyMRI segmentationneuromodulationneurostimulation using focused ultrasoundpeak pressureskull and soft tissue interactions in transcranial ultrasoundskull modelingtranscranial focused ultrasoundtranscranial ultrasound modelingultrasound beam distortion in brain tissuesultrasound safety and efficacy in brain treatments
