The same frequency, twenty to seventy times the intensity
The group behind the rejuvenation work also published the opposite experiment: low-frequency ultrasound that kills tumour cells while leaving normal cells untouched. Same transducer family, almost the same frequency, a very different dose. That comparison is the most useful calibration anchor this project has — and the clearest warning about getting the dose wrong.
What the cancer study actually did
Tijore et al., Ultrasound-mediated mechanical forces activate selective tumor cell apoptosis, Bioengineering & Translational Medicine 2025 (PMC11883105). Numbers below read from the paper's own text, not from a summary.
| Cancer study (Tijore 2025) | Rejuvenation study (Kureel 2025) | |
|---|---|---|
| Frequency | 33 kHz in vitro, 39 kHz in mice | 32.248 kHz |
| Reported dose | 39 mW/cm² in vitro (labelled "200 V"); mice at 39 / 51 / 166 mW/cm² (200 / 300 / 400 V) | "4 kPa" (or 4 Pa, depending on the version) |
| Duty cycle | 50% — 1 s on / 1 s off | 50% — 1.5 s on / 1.5 s off |
| Exposure | 2 hours in vitro; 1 h/day for 2–3 weeks in mice | 30 minutes |
| Set-up | Same device family: PZT4 rings with an aluminium cone, epoxy and silicone coated, bonded to the tank bottom; samples 8 cm above; degassed DI water; tank in a 37 °C incubator; signals generated by an Arduino | Transducer at the bottom of a 4 L beaker, sample 7–10 cm above, degassed water at 35 °C |
| Why degassed | Explicitly "to remove water microbubbles and prevent thermal heat generated by the microbubble cavitation effect"; the 50% duty cycle is described as a cool-off | Degassed, reason not stated |
| Target | Tumour cells on matrigel, patient-derived pancreatic organoids, chick CAM tumours, mouse CT26 subcutaneous tumours | Senescent fibroblasts, aged mice |
What it found, including the two bad results
The headline
- Apoptosis in tumour cells: 46% (MDA-MB-231 breast), 58% (A375p melanoma), 32% (HT1080 fibrosarcoma), against 9% / 4% / 5% untreated.
- Normal cells from the same tissue were unaffected: MCF10A 16% treated vs 15% control; HFF 3% vs 4%.
- Patient-derived pancreatic tumour organoids were killed.
- The selectivity has a mechanical explanation: the tumour lines tested lack rigidity sensors (tropomyosin 2.1, myosin IIA), so the same mechanical input produces a much larger calcium response.
The two results nobody quotes
- In mice, more power made tumours grow faster. Lowest tumour growth at 300 V (2–3 mm²); at 200 V it was 7–8.5 mm²; at 400 V it was about 10 mm², worse than the untreated controls at 4–5 mm². The authors' own conclusion: "the range power level for the killing of tumor cells is relatively narrow."
- Chick embryos died. Survival was 100% after the first exposure round, then fell to 60% after the second, against 100% in untreated embryos — described in the abstract as "relatively minor damage".
- The authors also flag the possibility of mechanically primed resistant cells and their role in metastasis as an open question needing in-depth in-vivo study.
The dose ladder: everything we know, in one scale
All published low-frequency results we have found, converted to peak pressure in water at ~33 kHz. This is the single most useful table on this site, and it is why calibration is not optional.
| Peak pressure | Intensity | What happens | Source |
|---|---|---|---|
| 2–5 kPa | ~0.1–0.8 mW/cm² | Stem cells spread, grow, differentiate toward bone (39 kHz) | IISc Bangalore 2026 |
| 4–8 kPa our target | 0.5–2 mW/cm² | Senescence reversal, autophagy up, SASP down (32.2 kHz, 30 min) | Kureel 2025 |
| 10–20 kPa | 3–13 mW/cm² | Kills normal stem cells (39 kHz) | IISc Bangalore 2026 |
| ≈15 kPa | 7.7 mW/cm² | Tumour killing enhanced after microtubule depolymerisation (33 kHz, 2 h) | Singh 2021 |
| ≈34 kPa | 39 mW/cm² | Tumour-cell apoptosis, normal cells unaffected (33 kHz, 2 h) | Tijore 2025 |
| ≈39 kPa | 51 mW/cm² | Best tumour suppression in mice (39 kHz, 1 h/day) | Tijore 2025 |
| ≈70 kPa | 166 mW/cm² | Tumours grew faster than untreated controls | Tijore 2025 |
| 50–75 kPa | ~85–190 mW/cm² | Apoptosis in oral cancer cells | IISc Bangalore 2026 |
| ~80–130 kPa | ~0.2–0.6 W/cm² | An ultrasonic cleaning bath at its lowest setting; inertial cavitation threshold in air-saturated water | measured / architecture A |
The whole span from "grows cells" to "kills tumour cells" to "makes tumours worse" is about 30× in pressure — roughly 30 dB. Our rejuvenation target and the tumour-apoptosis dose are separated by only 12–18 dB, which is less than the error of an uncalibrated DIY hydrophone (±3–6 dB) stacked on an unmapped field (up to 9× variation across a sample plane). An ultrasonic cleaner sits above the entire ladder.
Normalising the two protocols honestly
Converting with I = p²/2ρc for peak pressure and ρc = 1.5 × 10⁶ Rayl:
| Quantity | Rejuvenation | Cancer, in vitro | Ratio |
|---|---|---|---|
| Peak pressure | 4–8 kPa | ≈34 kPa | 4–8× |
| Intensity while on | 0.5–2 mW/cm² | 39 mW/cm² | 20–73× |
| Time with ultrasound on | 15 min | 60 min | 4× |
| Energy delivered per cm² | ≈0.5–2 J/cm² | ≈140 J/cm² | 70–290× |
Those ranges are wide for reasons that are not our fault: neither paper states whether its pressure is peak or RMS, nor whether the intensity is pulse-average or time-averaged, and the rejuvenation paper's instrument was a cavitation meter reporting a time-averaged total. A factor of 2 to 4 of the spread above is pure reporting ambiguity. Any replication that states its convention in SI units immediately does better than both papers.
One mechanism, two outcomes
Both results run through the same door: mechanical force opens PIEZO1, calcium enters, and what happens next depends on the cell and on how much calcium.
Senescent, non-malignant cell
Modest calcium → SIRT1 moves to the cytoplasm, mTORC1 down, autophagy and lysosomal activity up, mitochondria fragment and move more, SASP falls, the cell divides again.
Tumour cell lacking rigidity sensors
The same input produces a much larger calcium rise → calpain-1/2 activation → mitochondrial death pathway → apoptosis. Normal cells, which have intact rigidity sensing, do not get the large calcium transient.
What an analogous DIY project would mean
The honest answer has four parts, and only the first one is easy.
| Dimension | What changes | Verdict |
|---|---|---|
| Hardware | Almost nothing. Same 33 kHz transducer, same tank, same degassed water. You need roughly 5–8× the drive amplitude and a longer session. Our architecture B already covers both regimes; the cancer dose is still only a few watts electrical. | trivial |
| Measurement | Everything. The regimes are 12–18 dB apart, the response is non-monotonic, and overdosing made tumours grow. Without a calibrated hydrophone and a field map you cannot know which regime you are in — and at this dose the wrong answer is not "no effect". | now mandatory |
| Biology | A different project entirely. Cancer cell lines (MDA-MB-231, A375p, HT1080) are sold to institutions, not to individuals; you need matrigel, an incubator, annexin V staining, flow cytometry or imaging, and biosafety containment. Patient-derived organoids need ethics approval and a tissue source. | partner lab only |
| Humans and animals | Out of scope, permanently. Treating a tumour is a medical act; animal work needs ethics approval. The chick-embryo mortality and the dose-dependent tumour acceleration are precisely why. | no |
| Intellectual property | This is the one application where the landscape bites. US 12,714,885 is granted, and all of its independent claims are about immersion devices that trigger mechanically-induced apoptosis in tumour cells — i.e. exactly this experiment. Senescence reversal sits outside those claims; tumour apoptosis sits inside them. In Italy that family was never filed, so there is nothing to infringe here, but a US collaborator would be squarely inside it. | keep out of scope |
- This project does not build a cancer treatment device, does not claim one, and does not include tumour apoptosis in its stated purpose or in its firmware features. That is a scientific judgement and also the cheapest possible insurance against the only granted patent in the field.
- Nobody points this at a tumour, their own or anyone else's. Beyond the law, there is no imaging, no targeting, no tissue dosimetry, a non-monotonic dose-response where too much accelerated growth in mice, and an unanswered question from the authors themselves about mechanically primed resistant cells. Someone who delays real oncological treatment for a home ultrasound bath is the worst outcome this project could produce.
- Cancer-cell work happens only in vitro, only with a partner lab, and only as a dose-calibration and safety experiment — never as therapy development.
What we actually take from it
- A dose ceiling for our own device. Our acceptance test already caps the field; the ladder above tells us what lies just above it. Any build that cannot prove it is below ~10 kPa everywhere is not merely imprecise — it is in a regime that kills normal stem cells in vitro.
- A biological dosimeter. This is the most valuable idea on the page. The cancer response is sharply dose-dependent and published with numbers: tumour lines die at ≈34 kPa while matched normal lines do not. A partner lab running that assay against our rig tells us whether our "4 kPa" is really 4 kPa — a biological cross-check on the acoustic calibration, which is exactly what the original papers lack. The matched normal-cell line is the built-in negative control.
- A sharper design for the oncogenic-safety arm. We already planned to ask whether LFU wakes up dangerous senescent cells. The cancer branch adds the mirror question: at our rejuvenation dose, do transformed or pre-malignant cells get primed rather than killed? The Nature Communications result makes that a real hypothesis, not a hypothetical one.
- A reason to publish the convention, not just the number. Two papers from one group, four unit conventions, and a 70–290× uncertainty in delivered energy. Reporting pressure in SI units with the measurement convention stated is a contribution in itself.
- A question for the authors. Added to the open questions: how far apart are the two regimes on their own equipment, measured the same way, and has anyone run the dose series on a single cell population from rejuvenation through to apoptosis?
Sources: Tijore et al., Bioeng Transl Med 2025 · Singh et al. 2021 · Saigaonkar et al., ACS Nanosci Au 2026 · Capillary constrictions prime cancer cell tumorigenicity through PIEZO1, Nat Commun 2025 · Kureel et al., Aging Cell 2025.