Dental ultrasound

Why ultrasound: more precise where it matters, and no radiation

Dentistry adopted X-rays a century ago because nothing else could see through enamel. Peer-reviewed studies now show ultrasound measuring bone, gums and early decay to a fraction of a millimeter, with zero ionizing dose.

Camera vs ultrasound Overview How ultrasound works Why 3D Proof of concept Precision Radiation Limits

Side by side

What a camera sees. What ultrasound is designed to see.

What a phone camera sees: the tooth surface only — Phone camera

What ultrasound is designed to see below the surface — Ultrasound

Illustration. Phone-photo monitoring stops at the enamel surface; ultrasound returns an echo from every boundary beneath it.

How dental ultrasound works

Sound in, echoes out, a picture of what lies underneath

1 · A pulse of sound

A sensor sends a short, high-frequency pulse (20–60 MHz for teeth, lower for bone) into the tooth through a thin layer of gel or saliva. The frequencies are far above hearing and carry no ionizing energy.

2 · Every boundary echoes

Wherever the material changes, part of the pulse reflects: enamel to dentin, dentin to pulp, gum to bone, healthy enamel to a softened, demineralized patch. The time each echo takes back gives its depth; its strength says what it hit.

3 · Many angles become a volume

A single probe gives a 2D slice. Sensors on three sides of every tooth, firing together, give multiple views of the same structure, which software reconstructs into a 3D volume, the same principle used in industrial inspection and breast ultrasound tomography.

Cutaway of a tooth showing what ultrasound is designed to see below the surface

What X-rays see

Density. A radiograph is a shadow: dense tissue blocks the beam, soft tissue does not. That makes X-rays excellent for bone and large cavities, blind to gum thickness and soft-tissue inflammation, and late to early decay, which must lose 30–40% of its mineral before it shows on film.

What ultrasound sees

Boundaries and stiffness. Ultrasound reflects at every change of material, so it images gums and soft tissue directly, measures bone level and thickness against the gum, resolves enamel and dentin layers, and responds to early demineralization before it becomes a hole. It also reads through metal and ceramic restorations that blind X-rays.

What a phone photo sees

The surface only: alignment, color, visible plaque. Nothing under the enamel or below the gum line.

Watch it happen

One pulse, three echoes, one line of the image

Animated pulse-echo: a pulse leaves the sensor, reflects at the enamel, dentin and pulp boundaries, and three echoes return in order of depth

1 · Pulse

The sensor fires a microsecond burst of sound into the tooth through gel or saliva.

2 · Echoes

At each boundary (gel→enamel, enamel→dentin, dentin→pulp) part of the pulse reflects back. The bigger the change in material, the stronger the echo.

3 · Trace

The sensor records when each echo arrives. Time becomes depth; a row of these traces becomes a 2D image, and firing from many positions becomes a 3D volume.

Repeat a few thousand times a second across every sensor in the tray and the whole arch is captured in one acquisition.

Schematic animation. Layer thicknesses, timings and echo strengths are illustrative, not to scale. Animation pauses for visitors who prefer reduced motion.

Ultrasound, explained

How ultrasound works: from a vibrating crystal to a measured depth

Medical ultrasound has been used safely for more than 60 years. The physics is simple enough to fit on one page, and it is the same physics inside S-WAVE™.

1 · A crystal that talks and listens

A piezoelectric element changes shape when voltage is applied, producing a pulse of sound at millions of cycles per second (megahertz). The same element then listens: returning pressure waves squeeze it and generate a tiny voltage. One sensor is both loudspeaker and microphone, switching thousands of times a second.

2 · Sound reflects at every boundary

Each tissue has an acoustic impedance, its density times its speed of sound. Where impedance changes, part of the wave reflects and the rest continues. Gel to enamel is a big jump, so a strong echo; enamel to dentin, a smaller one; dentin to pulp, another. A softened, demineralized patch of enamel has a different impedance from healthy enamel, which is why early decay can return an echo before it is a hole.

3 · Time becomes depth

Sound travels at a known speed in each tissue (about 1,540 m/s in soft tissue, far faster in enamel). Measure how long an echo took to return, multiply by the speed, divide by two, and you have the depth of the boundary that produced it. A string of echoes along one line is an A-scan; sweep or array many lines and you have a 2D image (B-mode); stack 2D slices or fire from many positions and you have a 3D volume.

Frequency sets the trade-off

Frequency	Typical use	Resolution	Depth
2–5 MHz	Abdomen, heart, obstetrics	~0.5–1 mm	15–25 cm
7–15 MHz	Thyroid, breast, vessels, skin	~0.1–0.3 mm	2–6 cm
20–60 MHz	Eye, skin layers, teeth and gums (research)	~0.03–0.1 mm	a few mm to ~1 cm

Higher frequency means shorter wavelength and finer detail, but faster absorption and shallower reach. Teeth are small and near the surface, so dentistry can use the high end of the scale, which is where the sub-0.1 mm results in the research library come from.

What the echo also tells you

Strength: how different the two materials are, the basis for telling healthy from demineralized enamel, or gum from bone.
Shape: a crack or a rough surface scatters the pulse; a smooth boundary reflects it cleanly.
Shadowing: behind a strong reflector (a metal filling, dense bone) little sound continues, which is both a limit and a clue.
Speed of sound itself: measured across a known thickness it reports stiffness and mineral content, the principle of bone-density ultrasound.

Why it is safe to repeat

Ultrasound is mechanical vibration, not ionizing radiation. It deposits no dose and does not break chemical bonds. Its two safety limits are heating and cavitation, which diagnostic devices stay well below under FDA acoustic-output limits (thermal and mechanical index). That is why obstetric scans are repeated throughout pregnancy, and why a monthly dental scan is a reasonable design goal.

Speeds of sound and impedance values are typical literature figures for human tissues and vary with composition and temperature; frequency, resolution and depth ranges are approximate. Diagnostic-ultrasound safety limits refer to FDA acoustic-output guidance for cleared devices; S-WAVE™ must demonstrate compliance in its own testing.

Three dimensions

Why 3D ultrasound changes diagnosis

A 2D ultrasound image is one slice through the body, chosen and angled by whoever holds the probe. A 3D volume contains every slice at once. In the specialties that adopted it, three things changed: measurements, reproducibility and who can perform the scan.

Measurements become volumes

In 2D, a volume is estimated from a diameter and a formula. In 3D it is measured directly. Three-dimensional echocardiography measures left-ventricular volume and ejection fraction with less variability than 2D and agrees more closely with cardiac MRI, which is why guidelines prefer it when available. The same shift applies to a bone defect around a tooth or an implant: a volume, not a line on a radiograph.

Any plane, after the fact

With a volume stored, a clinician can cut it in any plane later, including planes a probe could never reach physically, such as the coronal plane of the uterus in 3D gynecologic ultrasound. For a tooth, that means the cheek-side, tongue-side and biting-surface views of the same structure, registered together, from one acquisition.

Less dependence on the operator

In 2D, image quality and measurement accuracy depend on probe angle and hand skill. Volume acquisition fixes the geometry, so the same scan can be taken by a less-trained operator and read by an expert elsewhere. Automated fetal-volume acquisition and AI-guided cardiac scanning by novices (FDA-authorized in 2020) are the medical precedents; S-WAVE™’s fixed tray is the dental version of the same idea.

Where 3D ultrasound is used today

Field	What 3D added
Obstetrics	Surface rendering of the fetal face and spine; standardized volume acquisition of the fetal heart and brain for later expert review.
Cardiology	True chamber volumes and valve anatomy; guidance of structural-heart procedures.
Breast imaging	Automated whole-breast 3D ultrasound for dense breasts; ultrasound tomography (SoftVue, FDA PMA 2021) rings the breast with transducers to reconstruct a full volume without compression or radiation.
Gynecology	Coronal-plane views of the uterus for congenital anomalies and IUD position.
Vascular and musculoskeletal	Plaque volume in carotid arteries; tendon and nerve mapping.
Image-guided surgery	Intra-operative volumes registered to CT/MRI plans, updated without radiation.

What this means for a mouth

Comparable scans. A fixed-geometry 3D volume taken this month can be registered to last month’s. Change, not a single snapshot, is the diagnostic signal: a 0.2 mm shift in bone level is invisible on one image and obvious across six.
Whole-arch context. One tooth’s bone loss is read against its neighbours and the opposing arch, as a periodontist does chairside.
Sub-surface and surface together. The same volume that holds the bone crest also holds the enamel surface, so tooth movement during aligner treatment and the bone around the moving root are in one data set.
Ready for AI. Volumes from a fixed tray are far more uniform than hand-held sweeps, which is what segmentation and change-detection models need to train on.
No dose budget. Serial imaging is the whole point, and only a non-ionizing method can be repeated monthly without a dose conversation.

ionizing dose, however many times the scan is repeated

sides of every tooth captured in one acquisition, by design

12×

a year, the monitoring cadence S-WAVE™ is designed for, versus twice-yearly X-rays

Medical examples describe FDA-cleared or approved products and published clinical practice in other specialties; they are not KronosMD products and do not predict S-WAVE™’s performance. 3D full-arch dental ultrasound in people has not been demonstrated; see Limits and open questions below. Statements about S-WAVE™ describe design intent.

Proof of concept in medicine

3D ultrasound has been in medicine for forty years. Ask any parent.

The 3D ultrasound picture of a baby’s face is the most familiar medical image in the world. It was first made in 1986, sold commercially in 1989 and became routine in the 1990s, because it is safe to repeat, the volume can be reviewed in any plane after the scan, and it shows surfaces and structure that a single 2D slice cannot. S-WAVE™ is designed to bring that same forty-year-old principle to teeth, gums and bone.

How 3D ultrasound grew up in obstetrics

1984 – 1986 · First 3D fetal images

Kazunori Baba (University of Tokyo) reports a 3D ultrasound system in 1984 and, in 1986, the first 3D images of a fetus, built by processing 2D scans on a mini-computer.

1989 · First commercial 3D scanner

Kretztechnik’s Combison 330 (Zipf, Austria) reaches the market: the first 3D ultrasound system sold to clinics.

1992 – 1993 · The fetal face, spine and brain in 3D

Kuo, Chang and Wu (Taiwan) image the fetal face, cerebellum and cervical vertebrae in 3D; Feichtinger (Vienna) shows 3D transvaginal images of 10-week embryos; Kretztechnik launches the Voluson 530D.

1995 · Surface rendering enters diagnosis

Merz (Mainz) demonstrates multiplanar, surface and transparent views for facial clefts and skeletal anomalies; Jurkovic (London) shows 3D accuracy for uterine anomalies.

1996 · Real-time 3D and 4D

Baba reports real-time 3D in The Lancet; two groups publish 4D fetal echocardiography; the first English-language 3D ultrasound textbook appears.

1997 – 1999 · A field forms

Matrix-array 3D probes become commercial; the First World Congress on 3D Ultrasound in Obstetrics and Gynecology is held in Mainz; ISUOG forms its 3D focus group.

Today · Routine, repeated, radiation-free

3D/4D fetal ultrasound is standard equipment in obstetric units worldwide, repeated through pregnancy because it deposits no ionizing dose. The same volume-imaging principle has since been cleared by FDA for the heart, breast and thyroid.

What the pregnancy precedent proves

Safety over decades. Pregnant women have been scanned, and rescanned, with 3D ultrasound since the 1980s; it is the modality chosen precisely because it is non-ionizing.
Volumes change diagnosis. Surface rendering made facial clefts and skeletal anomalies visible in a way 2D never did; multiplanar review let specialists read a scan taken by someone else.
From research to routine in a decade. First images 1986, first product 1989, world congress 1997, standard of care by the 2000s.
Hardware is the easy part now. Matrix-array probes and software reconstruction are mature; what S-WAVE™ adds is a fixed intraoral geometry and sensors tuned for enamel and bone.

What is different about a mouth

A fetus floats in fluid, the ideal acoustic window; teeth sit in air and saliva, so S-WAVE™ needs a tray that keeps every sensor coupled.
Obstetric probes work at 2–8 MHz to reach deep; teeth are shallow and need 20–60 MHz for sub-millimeter detail.
Enamel is the hardest tissue in the body and reflects strongly, which is both the signal and the challenge for seeing beneath it.
And the operator: an obstetric 3D scan is taken by a trained sonographer; S-WAVE™ is designed so the patient takes it at home.

The precedent keeps spreading. The same approach, 3D volumes reconstructed in software from standard ultrasound hardware with AI measurement, was FDA-cleared for the thyroid in September 2024 (PIUR tUS Infinity, 510(k) K240036), for whole-breast imaging, and for 3D echocardiography, which guidelines now prefer for heart-chamber volumes. Each is a specialty that replaced an estimate with a measured volume.

History per the clinical chronology at ob-ultrasound.net (Woo J.) and the cited primary reports; product and company names are trademarks of their owners, used for identification only, and none is affiliated with KronosMD. Obstetric and other medical 3D ultrasound devices are cleared products in other specialties; their safety and performance records do not predict S-WAVE™’s, which has not been tested in people. Statements about S-WAVE™ describe design intent.

Precision

Measured against the gold standards, to a fraction of a millimeter

The numbers below come from peer-reviewed studies that compared research-grade dental ultrasound with the reference method for each structure: micro-CT, CBCT, histology or physical probing.

Structure	Ultrasound result	Compared with	Study
Alveolar bone level	Within 0.07 mm; inter-rater ICC 0.97 (vs 0.69 for CBCT)	CBCT	Nguyen, Le et al., 2023
Alveolar bone, in vivo	Within 0.18–0.31 mm; bone thickness within 0.05 mm	Micro-CT	Wang, Le et al., 2026
Gingival (gum) thickness	Within ~0.2 mm, ICC up to 0.87, no probing	Invasive probing	Alizad-Rahvar et al., 2024
Enamel thickness	r = 0.97	Histology	OOOO, 2005
Early enamel lesions	Demineralized enamel separated; depth matched	Micro-CT	Kim et al., 2019
Caries in patients	82% sensitivity (bitewing X-ray: 75%)	Clinical reference	Matalon et al., 2007
Cracks under restorations	Cracked vs intact distinguished “unequivocally”	Histology	Culjat et al., 2005

0.07 mm

bone-level agreement with CBCT, with better reproducibility than CBCT itself

82% vs 75%

caries sensitivity, ultrasound vs bitewing X-ray, in 47 adults

~0.2 mm

gum thickness without a probe, in 1,360 sites

Studies used research-grade, probe-based ultrasound systems operated by clinicians, not S-WAVE™. They support the physics and clinical value of dental ultrasound and are not evidence of S-WAVE™’s performance. Summaries are KronosMD’s reading of the published abstracts.

Radiation

Dental X-rays are low dose. Ultrasound is no dose.

Dental radiographs are among the smallest medical exposures, and dentists manage them carefully under the ALARA principle. The point is not that X-rays are dangerous; it is that a monthly check cannot be done with them, and ultrasound removes the question entirely.

Bitewing X-ray5 µSv

Panoramic14–24 µSv

Full-mouth series (18 films)35–171 µSv

CBCT, small field≈100 µSv

CBCT, large field≈300 µSv

One day of natural background≈8.5 µSv

S-WAVE™ ultrasound0 µSv

Each dot = 5 µSv, about one bitewing. The dashed circle is zero: ultrasound is mechanical sound, not ionizing radiation.

Effective dose per examination; CBCT ranges vary widely by device and field of view.

Why this matters for monitoring

A bitewing every six months is fine. A bitewing every month, for years, for a child in aligners, is not something any dentist would prescribe. Monitoring between visits needs an imaging method you can repeat without limit, and only non-ionizing methods qualify.

What S-WAVE™ is designed to do with that freedom

Scan every month, compare each scan with the last, and alert a dentist only when something changes. No dose budget, no office visit, no operator.

Dose figures: Health Physics Society summary of Ludlow et al. 2008 (bitewing 5 µSv; panoramic 14–24 µSv; full-mouth series 35–171 µSv); NCRP natural background ~3,100 µSv/year; CBCT ranges per published surveys (Ludlow et al. 2015) vary from tens to hundreds of µSv by field of view. Ultrasound is non-ionizing; diagnostic ultrasound carries its own safety limits on acoustic output, which S-WAVE™ testing must meet.

Limits and open questions

What the research does not yet show, and how we plan to test it

No study has imaged a full arch in 3D in a person

Every 3D attempt in the literature used a single swept probe on a phantom, a few teeth or an implant model. The bench prototype’s full-arch surface is, to our knowledge, rarer than anything published, and it has not been reproduced in a mouth.

Consumer use is untested

Published results come from clinicians operating research systems. Whether a lay user can seat a tray and obtain a usable scan is a human-factors question that testing must answer.

Saliva, air and movement

Ultrasound needs continuous contact. A fixed tray with a conformable couplant is the design answer; its performance across mouths, braces and restorations has to be demonstrated.

AI accuracy

The scan-comparison software does not exist yet. Its standalone performance and its clinical accuracy against reference imaging are part of the FDA testing plan.

Planned validation path

Bench

Accuracy of the prototype against micro-CT on extracted teeth and models.

Pre-Submission

Ask FDA to confirm the pathway and the testing plan (Q1 2027 target).

Verification

Acoustic output, biocompatibility, software V&V, AI standalone performance, lay-user human factors.

Clinical accuracy

S-WAVE™ scans compared with reference imaging in people, as FDA requires.

A peer-reviewed paper is planned. KronosMD intends to publish its bench research and scan data in a PubMed-indexed dental journal, so the claims on this page can be checked by people who do not work for us.

S-WAVE™ is in development, has not been tested in people and is not cleared or approved by the FDA. Target dates are estimates and may be delayed; the FDA could require more testing than we expect, or may never authorize S-WAVE™.

Dental imaging without radiation, from home.

Questions about S-WAVE™, the research or a collaboration? Our team is here to help.