Skip to content
No. 039OpenRadiation oncologyUpdated Nov 16, 2026

Can the low-power 'modulated electro-hyperthermia' device that a South African randomised trial added to chemoradiotherapy for cervical cancer (planned at 90 to 130 watts) actually raise the temperature of a deep pelvic tumour enough to sensitise it to radiation, judged by patient-specific electromagnetic and heat simulations across many real pelvic anatomies?

Good it would do
2 / 5
Can AI help
3 / 5
How neglected
3 / 5
Can we check it
2 / 5

Why it's stuck

The device is promoted for low-resource settings, and the South African trial (Minnaar and colleagues, PLoS One 2019, 210 women randomised, many HIV-positive) reported higher six-month local disease control with it (45.5% against 24.1%, p = 0.003). The trial's authors say the device costs less than half the price of a brachytherapy device and is 'suitable for investigating as a chemo-radiosensitiser in a low-resource setting'.

What we'd try

Build anatomical models of the female pelvis from public segmented scans (for example the TotalSegmentator MRI dataset on Zenodo, 1,296 scans with 50 segmented structures, CC-BY 4.0, and its CT counterpart), place a cervical tumour of realistic size, model the device's capacitive electrodes and water bolus at 13.56 MHz, and solve the electro-quasistatic field and the Pennes bioheat equation for each anatomy at 90, 110 and 130 W. Report the distribution across patients of the tumour temperature reached, and the thermal dose (CEM43), with tissue-property and perfusion uncertainty propagated by Monte Carlo sampling.

What could stop us

The answer may already be settled by an earlier analytical physics paper and by existing capacitive-heating simulations, and the only patient temperature data to check a simulation against are one small manufacturer-reported study.

Source: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0217894

Do you know the answer?

One reply from someone in the field can settle what our AI can't.

Members can read the full record for this problem, check its sources and comment. Open the member view → Not a member? Here is how to join.