Microwave and millimeter-wave sensing has become one of the most promising non-invasive routes to skin cancer diagnosis. The technique exploits a simple physical fact: malignant lesions such as basal cell carcinoma, squamous cell carcinoma, and melanoma contain more water than healthy skin, and water-rich tissue has a higher dielectric permittivity at microwave frequencies. By placing a transmitter and a receiver antenna in contact with the skin and exciting electromagnetic surface waves that travel along the skin-air interface, the system can detect the extra attenuation and phase shift introduced by a tumor. The approach is non-ionizing, painless, low-cost, and capable of real-time readout, and recent advances in antenna design and deep-learning signal processing have pushed spatial resolution and diagnostic reliability steadily upward.

Yet safety has lagged behind performance. Most previous evaluations of skin-cancer detectors relied on a single-tissue model representing a single anatomical site, typically the forearm or abdomen, and most assessed safety solely through the specific absorption rate, or SAR, the standard metric that quantifies how much electromagnetic energy is dissipated per kilogram of tissue. The McGill team argues that both shortcuts are problematic. Anatomical differences in skin and subcutaneous structure across the body strongly influence the electric-field distribution at microwave frequencies, so a single-site assessment may not reflect the actual risk during clinical deployment. And SAR itself, while codified in regulatory frameworks, does not capture exposure duration, tissue perfusion, or heat diffusion, all of which determine the physiological outcome that actually matters: how much the tissue warms up.

The two antennas under study were both built on 50-micrometer-thick flexible Kapton substrates chosen for biocompatibility and good skin contact. The monopole antenna, the team’s first-generation design, measures 20 by 20 millimeters and operates from 2 to 5 gigahertz, with a standard SMA connector on the reverse side for repeatable measurements. The newer Vivaldi antenna, measuring 26 by 17 millimeters and operating from 3 to 6 gigahertz, was selected for its end-fire radiation pattern and wide bandwidth, which confine energy along the skin surface. Because biological tissue has a high permittivity, both antennas are far smaller than their free-space counterparts, a key advantage for wearable and point-of-care deployment.

To capture anatomical reality, the researchers constructed nine multilayer tissue models in Ansys HFSS, each consisting of skin, fat, muscle, and, where relevant, bone. Layer thicknesses came from published anatomical measurements: skin thicknesses ranged from 0.9 millimeters on the forehead to 1.7 millimeters on the buttock, while fat thickness varied from a mere 0.3 millimeters on the palm to 12.1 millimeters on the buttock. For the forehead, palm, and sole, all four layers were included because soft tissue is thin enough for the waves to reach bone. At the other six sites, where soft tissue exceeds 20 millimeters, the bone layer was omitted, since microwave penetration depth in soft tissue at 2 to 6 gigahertz is less than 20 millimeters and the waves attenuate before reaching bone. Each model shared identical lateral dimensions of 12 by 6 centimeters, large enough to accommodate the compact antennas.

The full-wave simulations revealed a consistent pattern for both antennas: the peak electric field, and therefore the SAR hot spot, always appears in a small area on the skin surface directly beneath the transmitting antenna. But the magnitude of that hot spot varies enormously with anatomy. The forehead model, with only about 5.1 millimeters of soft tissue over bone, produced the strongest fields, because the wave reflects strongly at the bone interface and constructively reinforces the field at the surface. Sites with much thicker soft tissue, roughly 11 to 33 millimeters, showed far weaker surface fields. At 6 gigahertz, the peak SAR in the forehead model approached nearly twice that of the sole or neck models at identical input power, and the monopole results showed the same trend, with all other sites registering 43 to 54 percent lower peak SAR than the forehead.

Frequency matters as well. Across the operating bands, peak SAR rose steadily with frequency in every tissue model. The explanation lies in the dispersive nature of tissue dielectric properties: as frequency increases, conductivity rises while permittivity falls, which strengthens the skin effect and confines the electromagnetic field ever more tightly to the superficial layers. Because SAR is proportional to the square of the local electric field times conductivity, the combination of higher conductivity and a more surface-localized field drives peak absorption upward at higher frequencies.

The temperature analysis, performed by solving Pennes’ bioheat equation with an explicit finite-difference scheme in MATLAB, added a twist that SAR alone could never reveal. Under a 10-dBm Vivaldi scan over 100 seconds, the palm model heated the most, reaching a peak rise of 2.47 degrees Celsius, while the buttock model, despite having the second-highest peak SAR of all nine sites, produced the smallest temperature rise, about 20 percent lower. The ranking of temperature rise simply does not follow the ranking of SAR, because heating depends not only on the absorption hot spot but also on the spatial and temporal distribution of deposited power, the thermal properties of each layer, and heat diffusion. Even in a short 10-second exposure, comparable to a real clinical scan, the differences persisted: the arm model showed the highest rise at 0.516 degrees Celsius, roughly 16 percent above the buttock. The authors note that at most sites the 1-degree-Celsius safety threshold adopted by IEEE Standard C95.1 would be reached within 30 seconds at 10 dBm, meaning that power level is simply too high for practical use.

From the SAR results, the team derived site-specific power limits that keep the peak spatial-average SAR below the 1.6 watts-per-kilogram threshold mandated by both the U.S. FCC guidelines and Health Canada’s Safety Code 6, averaged over 1 gram of tissue. For the Vivaldi antenna, allowable power ranged from as low as 2.2 dBm at the forehead up to 4.9 dBm at the sole, so the researchers adopted the strictest value, 2.2 dBm, as the global limit for full-body use. Thermal verification confirmed the conservativeness of that choice: after a 10-second exposure at 2.2 dBm, the highest temperature rise anywhere was only about 0.068 degrees Celsius, negligible by any clinical standard. For the monopole antenna, the corresponding global limit was 2.9 dBm, again set by the forehead, and the worst-case 10-second temperature rise at that power stayed below 0.09 degrees Celsius, occurring at the palm.

The authors are careful to note the limitations of their thermal model, which omits blood perfusion, metabolic heat, and convective cooling at the skin-air interface, making the temperature predictions deliberately conservative upper bounds. Future work will incorporate perfusion and heterogeneous boundary effects. Even so, the study delivers something the field has lacked: a standardized workflow, from electric-field mapping and SAR calculation to temperature-rise evaluation and power-limit derivation, that bridges electromagnetic design and biomedical safety. For a technology racing toward clinical integration, the message is clear. A safety assessment based on a single body site is not a safety assessment at all, and the device that scans your forehead may need very different power settings from the one that scans your sole.

Subject of Research: Radiation safety analysis of microwave surface-wave antennas for skin cancer detection across nine anatomical sites

Article Title: Site-dependent numerical safety analysis in microwave skin spectroscopy: comparison of monopole and vivaldi antennas

Article References: Shang, S., & Popović, M. (2025). Site-dependent numerical safety analysis in microwave skin spectroscopy: comparison of monopole and vivaldi antennas. Frontiers in Antennas and Propagation, 3, Article 1714578. https://doi.org/10.3389/fanpr.2025.1714578

Image Credits: AI Generated

DOI: 10.3389/fanpr.2025.1714578

Keywords: microwave sensing, skin cancer detection, specific absorption rate, antenna safety, temperature rise, bioheat equation, Vivaldi antenna, monopole antenna, wearable devices, electromagnetic simulation, radiation exposure, tissue modeling