Maarten Paulides is a professor of (bio)medical electromagnetics at Eindhoven University of Technology, chair of the Electromagnetics for Care & Cure Lab (EM4C&C) and scientific director of the Center for Care & Cure Technology Eindhoven (C3TE).

25 September 2023

The distinct features and non-invasive nature of electromagnetic waves at different frequencies place them at the core of many medical applications. TUE’s Maarten Paulides gives an overview of the field.

Improvements in hygiene, diet, disease prevention and healthcare have steadily increased our life expectancy. Our aging society now poses the question of how to balance improvements within the available health budget. The number of chronic and age-related diseases is also increasing and ever older and more fragile patients are being treated. In this landscape, how can we provide broad access to the appropriate cure and care?

Medical technology (medtech) is expected to play an important role in the balancing act by providing cost-effective monitoring, diagnosis and therapy. Especially, electromagnetic (EM)-based approaches will form a cornerstone in our answer to the societal need for high-quality sustainable healthcare. The distinct features and non-invasive nature of electromagnetic waves at different frequencies place them at the core of many medical applications.


At sub-kHz frequencies, electric conduction dominates the EM effects in tissue. Important applications at low frequencies are neuromonitoring and neuromodulation. These are based on the currents that are used to communicate internally via neuronal pathways within the brain, as well as with the rest of the body. Using electroencephalography (EEG), we measure which parts of the brain are active and what state of activity they’re in.

Using EM waves below 100 Hz, the brain can also be stimulated to treat neuronal conditions like epilepsy and Parkinson’s, but also to relieve pain. Maybe the most established option is stimulation using electrodes inserted by deep brain surgery (deep brain stimulation, DBS). EM waves, however, also allow a more friendly option by stimulation from outside the skull like using transcranial magnetic stimulation (TMS). At TUE, in collaboration with Gent University Hospital, we’re researching how direct currents from outside can be applied to treat epilepsy using transcranial direct current stimulation (TDCS).

Bits&Chips event sponsor registration

One of our findings in a volunteer study however suggests that the TDCS currents might not reach as deeply as commonly assumed. Therefore, we started to investigate the recently discovered temporal interference (TI) technique. This approach exploits two electrode pairs driven at frequencies in the kHz range but with a differential frequency under 30 Hz. The neurological response acts on the differential frequency because it sort of operates like the diode in a radio that extracts the low-frequency audio signals from the amplitude-modulated (AM) waves propagating through the air. The electrode pairs can be placed at different sides of the head, to tailor where the two signals interfere. Hence, pre-intervention planning using advanced neuro-EM modeling would enable to position the interference region at the desired location.

While TI is very promising, more research is urgently needed to develop devices that act deeper and more directly on the intended bioelectric effect. Bio-EM modeling is also imperative to pre-plan and support clinical research to elucidate which diseases or disorders can be treated using TI, offset against alternatives like TDCS. It’s important to note that safe and reliable operation may also mean that integration with brain monitoring (MRI, EEG) will be required, eg to enable feedback-controlled neuromodulation.


Higher frequencies in the MHz range, ie RF waves, enable applications for non-contact diagnosis, monitoring and hyperthermia. Important EM devices, used en masse around the globe for imaging, are (nuclear) magnetic resonance (MR) scanners. These exploit an extremely strong magnetic field (B0) to align the normally randomly aligned spinning of atoms like hydrogen. Additional energy, in the form of an RF wave at the right (Larmor) frequency and in the right direction, deflects the spin orientations to their higher energy state. After RF power is stopped, the spin orientations gradually return to their lower energy state, while emitting an RF frequency that can be measured. Crucial for MR imaging is that both B0 and transmitted RF are as homogeneous as possible to avoid (strongly non-linear) imaging inhomogeneities.

Clinically used field strengths usually operate at 1.5-3 T, corresponding to RF between 64-128 MHz, respectively. Because both the Larmor frequency and the achievable signal-to-noise rate (SNR) depend on B0, research devices exploit field strengths at or above 7 T. Unfortunately, this brings the Larmor frequency to 298 MHz and above, making the RF wavelength similar to or smaller than the human body, leading to unwanted black regions due to wave interference of the transmitted RF signals. Therefore, various means for homogenization of the excitation field are under study.

The ‘lens’ of the MR scanner is the RF coil measuring the received signals. Since coils are sensitive in the near field, they should be positioned as closely to the region of interest as possible. Hence, advanced coil and coil-array configurations are under study that allow for a more homogeneous signal reception while keeping unwanted RF heating low, eg using flexible electronics to conform to the patient’s skin.

Another important challenge of MRI is safety, specifically at higher field strengths, if implanted metal devices or wires are present inside the patient or when therapy devices are used inside the scanner. Dedicated research is being conducted into bio-EM models to predict tissue heating and assess the risks related to new devices or protocols.


Biological evidence and clinical studies provide compelling evidence that moderate heating (39-44 °C) of tumor tissue is a strong sensitizer for both radio and chemotherapy. Unfortunately, widespread exploitation of the biological effects of hyperthermia therapy is impeded by the challenges of targeting the heating: the ability to heat only the tumor area and the monitoring to steer the heat.

In EM hyperthermia, choosing the right frequency is an important factor since the EM absorption in tissue and the heating precision are both strongly frequency dependent. The higher the frequency, the higher the absorption of EM waves in tissue, so the lower the penetration. Circular phased arrays of EM antennas can heat 5-10 times deeper and confine the heating – the radius of the EM focus of a circular array is one-third of the wavelength in tissue.

The MRcollar is a new device for MRI-guided hyperthermia.

Clinical phased-array devices employ 12 to 24 sources at 70-140 MHz, while those in the research phase generally operate at 433 MHz or beyond. Another challenge is accurate monitoring of the heating. Modeling thermoregulation in tissue is extremely complex. Hence, while modeling is increasingly used for pre-planning and guidance during the treatment, temperatures must still be measured using probes placed in tissue or body cavities. This approach is cumbersome and provides information only at the probing locations.

Research is directed at measuring temperature non-invasively using MRI. For this, devices were developed that work inside an MR scanner. While such MR thermometry provides the option to measure temperature in 3D, the challenge of this approach is currently to overcome the strong impacts of motion, like breathing and moving gas in the intestines, on the measurement.

Thermal ablation

Above 2.5-5 GHz, penetration of EM waves is too low for therapy beyond the skin. Still, ample opportunities exist here for (minimally) invasive applications, like thermal ablation, or applications that require less power transfer to the region of interest, like wireless communication with (medical) implants.

Radiofrequency ablation (RFA) exploits EM heating around a needle pierced into tissue or a catheter applied through a vessel nearby. RFA heating is used to cure cancers (lung, liver, kidney and bone), block abnormal electrical pathways in the heart, close veins during surgery and destroy areas of the nerve for relief of chronic pain. Loosely speaking, it’s like boiling an egg. The effect is time and temperature dependent.

Challenges in RFA include the precise control of the ablated region, especially for large volumes. Overheating superficial tissue leads to charring, reducing the wave penetration. Air bubbles and cooling by blood vessels also reduce the ablated volume. Since MR thermometry is usually impossible, research is directed at easy-to-apply ways of temperature monitoring that still permit accessibility, and detailed uncertainty assessments using modeling. Animal testing currently is the main method for demonstrating effectiveness and safety.

In the Dutch NeurotechNL consortium, we aim to exploit GHz waves to communicate with brain-implanted devices and low-MHz waves to power them. One specific example is a vision-restoration device that uses many implanted chips to communicate with the neuronal cells in the visual cortex. To avoid infections, this should be done wirelessly. The application requires a high data rate, for which we research options to apply wideband techniques for communication to and from implants, which need to be as small as possible. The small form factor also is a challenge for wireless power transfer, since efficient transfer to small implants is very complex. In this application, too, animal testing is imperative for demonstrating effectiveness and safety.

Wireless power transfer
Efficient wireless power transfer to small implants is very complex.

Bio-EM modeling

Computational modeling is transforming the path to market for medical devices. Regulatory commissions throughout the world have started to recognize the great strength of computational modeling and digital evidence in medical device development and assessment. Meanwhile, the reinforcement of the new European Medical Device Regulation (EU MDR 2017/745) requires an increased number of studies even at the prototype level to meet new European standards.

These new provisions reinforce the quality of European medical devices. However, the higher legal requirements at each development step also tremendously increase the development costs. And as certification bodies are overloaded, device re-certification is currently stagnating. At the same time, MDR-compliant design of new devices is seeing an exponential increase in preparation time and costs for early clinical trials required for each consecutive device iteration.

An important contribution to removing this hurdle is by applying virtual design based on accurate “non-animal human-data-based” (NAHB) patient modeling. NAHB simulation tools allow for much faster, and therefore cost-effective, device development. They’re starting to reduce the iterations needed for testing interim designs. Such simulators also provide a non-animal way of proving devices are safe.

This approach requires (a lot of) standardized measurement data acquired during patient treatments, preferably stored in open-access repositories. Obtaining, storing and analyzing this wealth of information crucially calls for collaboration in the “Golden Triangle” in medtech – industry, hospitals and academia. Co-development of such models in this triangle is imperative to ensure the connection between engineering research advances, biological and clinical knowledge and the needs of society.

Main picture: using transcranial direct current stimulation (TDCS), direct currents from outside can be applied to treat epilepsy.

Edited by Nieke Roos