Whole Cross-Sectional Ultrasound Tomography: Evidence, Limits & SoftVue
“Whole-body ultrasound” can be misleading. The peer-reviewed Caltech work demonstrated complete two-dimensional cross-sections of the abdomen and thighs in five healthy adults—not a clinically validated head-to-toe screening examination. Midjourney Medical has separately publicized a full-body prototype, but its speed, resolution, diagnostic performance, cost, safety profile for repeated screening, and deployment schedule have not been established in peer-reviewed clinical studies.
A useful commercial reality check is SoftVue. It also uses a water bath and a transducer ring, but it images one breast at a time. FDA approved SoftVue in 2021 for a specific adjunctive dense-breast screening indication; that authorization does not extend to whole-body scanning.
Why Water?
The large acoustic-impedance mismatch between air and tissue causes nearly total reflection at an air–skin boundary. Gel removes small air gaps for handheld ultrasound; a water bath provides continuous coupling between a surrounding array and immersed anatomy. Water's impedance is close to that of soft tissue, so reflection at the water–tissue interface is relatively small. The relevant benefit is coupling, not simply that sound travels faster in water than in air.
A ring geometry can collect many angular views, but the Caltech instrument did not fire every element into the body simultaneously: the paper describes a 512-element circular receiver array and a rotating transmitter. Water also does not solve fundamental propagation barriers. Gas-filled lung and bowel and mineralized bone strongly limit or distort ultrasound, so this technique cannot be assumed to reproduce CT lung imaging, CT bone detail, or MRI brain imaging.
What the Caltech Study Actually Showed
Published: April 24, 2026 in Nature Biomedical Engineering
Institutions: Caltech and City of Hope National Medical Center
Participants sat in a vertical immersion tank with the head above water. The system used a custom 512-element circular receiver array and a rotating transmitter; vertical positioning selected the slice level. It acquired reflection and transmission data used to reconstruct the following image types:
| Parameter | What It Measures | Clinical Value |
|---|---|---|
| Echo (Reflection) | Standard ultrasound echoes at tissue boundaries | Structural contrast from acoustic interfaces |
| Speed of Sound | How fast sound travels through different tissues | Quantitative tissue-property contrast |
| Attenuation | How much sound energy is absorbed or scattered | Energy-loss contrast; not a direct stiffness measurement |
The feasibility study included five healthy adults and imaged abdominal and thigh cross-sections. One two-dimensional slice required about 10 seconds of acquisition; that is not a demonstrated head-to-toe examination time. Sequential UST and 3-T MRI images showed anatomical agreement, but the study did not test sensitivity, specificity, reader performance, outcomes, or noninferiority to MRI or CT.
The paper also explored adipose-layer measurement and biopsy-needle localization. The authors reported a planned collaboration with City of Hope to evaluate liposarcoma. A plan for patient research should not be described as an established clinical indication or as proof that the system can detect cancer.
SoftVue: An FDA-Approved—but Breast-Only—Comparator
SoftVue is an automated whole-breast ultrasound tomography system made by Delphinus Medical Technologies. The patient lies prone with one breast suspended in a warm water bath. A transducer ring moves vertically and acquires coronal slices using reflection, sound-speed, and attenuation information.
FDA PMA P200040, approved October 6, 2021, indicates SoftVue as an adjunct to mammography for screening asymptomatic women after mammography confirms BI-RADS c (heterogeneously dense) or d (extremely dense) breasts. Images are interpreted with the mammogram. It is explicitly not a replacement for screening mammography. Approval permits marketing for that indication; it does not mean the device is installed at every breast center, so local access must be confirmed with a provider.
The FDA evidence included prospective case collection followed by multireader, multicase retrospective reader studies. In the pivotal RRS3 set (140 selected cases: 36 cancers and 104 noncancers; 32 readers), the supportive lesion-localization analysis found an absolute AUC increase of 0.0548 and, at a BI-RADS 4 threshold, an average sensitivity increase of 0.0738 with a specificity decrease of 0.0526. These enriched reader-study results are not population screening rates and should not be converted into an unqualified “percent more cancers” claim.
USC Keck School of Medicine was one of seven case-collection sites listed in FDA's RRS2/3 data. USC did not develop the Caltech device; the Caltech paper's clinical coauthor and planned patient collaboration are from City of Hope. Keeping those institutional roles separate avoids a common conflation.
Midjourney's “Ultrasonic CT”: Prototype Claims, Not Clinical Evidence
Midjourney Medical announced a water-immersion, ring-array prototype in June 2026 and calls it “Ultrasonic CT.” The name describes tomographic reconstruction; the system does not use x-rays and is not a CT scanner in the conventional radiology sense. Company materials describe Butterfly Network ultrasound-on-chip components, a future sub-60-second scan, body-composition output, and spa deployment. Those remain company statements rather than independently replicated clinical findings.
- No published diagnostic-performance study: no peer-reviewed sensitivity, specificity, cancer-detection, false-positive, or outcomes evidence was identified.
- No authorization for this system: clearance or approval of a component does not authorize a new assembled scanner or a screening claim.
- Targets are not current specifications: a 60-second scan, submillimetre detail, future sites, volumes, and costs require independent verification.
- Promotional images are ambiguous evidence: renders and segmented visualizations are not substitutes for raw clinical images, methods, reader studies, or regulatory labeling.
Evidence-Based Comparison
| Feature | Caltech investigational UST | SoftVue | Midjourney prototype |
|---|---|---|---|
| Anatomy studied/claimed | 2D abdomen and thigh cross-sections | One breast at a time | Full body claimed |
| Human evidence | Feasibility in 5 healthy adults | FDA-reviewed reader-study evidence | No peer-reviewed clinical validation identified |
| Acquisition | 512-element receiver ring plus rotating transmitter | 3-MHz transducer ring moving through a breast water bath | Ring-array architecture described by company |
| Timing | ~10 s per 2D slice—not whole-body time | Do not infer from Caltech timing | <60 s is a target, not validated exam time |
| Regulatory status | Investigational; no FDA authorization identified | FDA PMA approved in 2021 for narrow adjunctive indication | No FDA authorization identified |
| What it can replace | No replacement claim established | Explicitly not mammography | No replacement claim established |
Safety, Screening, Cost, and Modality Limits
- Safety: Diagnostic ultrasound is non-ionizing and generally regarded as safe, but it can produce thermal and mechanical bioeffects. Use should be medically justified and exposure kept as low as reasonably achievable; “entirely harmless” and unlimited-frequency claims are too absolute.
- Whole-body screening: No evidence cited here shows that screening asymptomatic people with whole-body UST improves health outcomes. Incidental findings, false positives, false negatives, downstream tests, anxiety, and overdiagnosis must be studied.
- Dense breasts: SoftVue's FDA approval establishes reasonable assurance of safety and effectiveness for its labeled adjunctive use. Separately, the 2024 USPSTF found evidence insufficient to recommend for or against supplemental ultrasound or MRI after an otherwise negative mammogram in women with dense breasts. FDA authorization and population-level guideline recommendations answer different questions.
- CT and MRI: CT remains strong for lung, bone, acute hemorrhage, and many rapid whole-body indications; MRI offers diverse soft-tissue, neurologic, vascular, diffusion, and contrast mechanisms. UST cannot be called equivalent on the basis of visually similar examples.
- Cost and time: Caltech's study did not establish a reimbursed price, cost-effectiveness, complete examination duration, interpretation time, or workflow. “Fraction of CT/MRI cost” and “much faster than MRI” are projections unless demonstrated for a defined clinical task.
The Bottom Line
- Caltech demonstrated reflection/transmission UST of full cross-sections, not a validated whole-body screening exam.
- The evidence base was five healthy adults; no diagnostic-performance or CT/MRI equivalence claim is justified.
- SoftVue is clinically authorized, but only for breast imaging alongside mammography in the labeled dense-breast population.
- Midjourney's timing, resolution, clinical utility, price, and deployment remain prototype or company projections.
- Water improves coupling; it does not remove the physics limitations of bone, gas, attenuation, and refraction.