Pulp vitality assessment is essential in dental diagnostics, yet current clinical methods remain unreliable. Microwave sensing offers a non-invasive alternative by detecting changes in the electromagnetic properties of pulp tissue, but these dielectric properties are not well documented. This study characterizes the dielectric behavior of primary and permanent bovine dental pulp, selected for its similarity to human pulp, across 1-20 GHz, including temporal changes that model necrotic tissue and are supported by histological validation. Dielectric measurements were collected from 22 freshly extracted bovine incisors (12 permanent and 10 primary) using an open-ended coaxial probe. Analysis shows no significant differences in dielectric properties among coronal, middle, and apical regions in either tooth type, indicating structural uniformity, while primary teeth exhibited lower dielectric constants than permanent teeth (P < 0.05). Temporal analysis revealed a significant decrease in dielectric constant from fresh to necrotic tissue, particularly at low frequencies. Cole-Davidson formulations were utilized to accurately represent the dielectric response across the investigated pulpal conditions, with first- and second-order models providing the best agreement depending on the dielectric complexity of the tissue state. These findings provide fundamental dielectric data for dental pulp and establish a quantitative basis for developing microwave-based, non-invasive pulp vitality assessment systems.
Typical microwave breast imaging systems use 8-72 antennas, which controls hardware complexity. This paper investigates whether a compact four-antenna array and machine learning can localize a tumor surrogate directly from measured S-parameters, without image reconstruction. Four UWB monopoles antennas around a 3D-printed breast phantom measure 16 complex S-parameters from 0.1 to 8 GHz. After a single preprocessing step, subtraction of a start-of-session baseline, a convolutional neural network (CNN) classifies each 32 × 791 magnitude-and-phase matrix directly to a discrete tumor grid position. Performance is evaluated with strict leave-one-session-out cross-validation on phantoms with and without fibroglandular inserts and, on the adipose-only phantom, across drift-test scenarios including VNA recalibration, antenna reattachment, antenna swap, canola-oil replacement, and cross-day measurement. On the adipose-only phantom the CNN reaches 100% per-position accuracy and maintains 96-100% under every perturbation; accuracy is 92% with the medium glandular insert and 67% with the largest, tracking the measured tumor-induced signal. Ablations show the adipose-only phantom tolerates two antennas (99.3%) and a 2-5 GHz band (98%), while the glandular phantoms require the full array and bandwidth, with their information carried primarily by the phase response. Four of five classical classifiers match the CNN at the full array. Because the preprocessing requires no session-level statistics, the trained pipeline classifies individual live measurements in approximately one second, supporting a path toward low-cost point-of-care breast screening.
At millimeter-wave frequencies used in fifth-generation (5G) wireless systems, electromagnetic absorption is mainly confined to the skin. This makes the way skin is represented in numerical models particularly important for exposure assessment and computational dosimetry. Simplified multilayer models are still widely used, and the viable epidermis and dermis are often assigned the same dielectric properties, although they differ in composition and water content. In this study, we investigated how anatomical interface morphology and hydration-related dielectric contrast affect the electric-field distribution in human skin exposed to a normally incident plane wave at 26 GHz. Three simulation cases were considered: an idealized planar skin model, an anatomically realistic model with identical dielectric properties assigned to the viable epidermis and dermis, and a hydration-adjusted realistic model in which the viable epidermis properties were modified to account for its lower water content. Compared with the planar model, the anatomically realistic model produced localized field redistribution at the stratum corneum–viable epidermis and dermis–subcutaneous adipose tissue interfaces, with field variability increasing by 5.5- and 8.4-fold, respectively. The hydration-adjusted case mainly affected the viable epidermis–dermis interface, where field variability increased by 23% and the upper tail of the distribution shifted toward higher values. These results indicate that, even when the overall attenuation trend is preserved, anatomical and dielectric heterogeneity can modify local exposure patterns at skin interfaces. The proposed modeling approach may help refine skin dosimetry at millimeter-wave frequencies and support future mesoscopic analyses of localized electromagnetic interactions in superficial tissues.
Reliable wireless communication is one of the major requirements for modern scalp-implantable biomedical devices operating inside the human body. This work presents a miniaturized circularly polarized antenna designed for 915 MHz for reliable in-body biotelemetry links. The antenna is implemented on a Rogers RO3010 substrate with a compact footprint of $4 \times 4 \times \text{0.13}\,\text{mm}^{3}$. A modified meandered radiator together with an optimized slotted ground configuration is utilized to improve impedance matching and circular polarization while maintaining a highly compact profile. The antenna characteristics are investigated inside realistic scalp-tissue environments and validated experimentally using tissue-equivalent phantom models. Remarkably, the proposed antenna achieves impedance bandwidth of 32.78% (0.76 to 1.06 GHz), an axial-ratio bandwidth of 53.55%, and a realized gain of $-29.2$ dBiC at 915 MHz. In addition, safety analysis confirms compliance with IEEE specific absorption rate requirements, whereas the link-budget evaluation demonstrates stable wireless telemetry beyond 10 m. The combination of small size, broad axial-ratio bandwidth, and reliable in-body communication makes the proposed antenna suitable for next-generation medical healthcare systems.
Microwave breast imaging seeks to detect lesion-related electromagnetic scattering without ionising radiation, but weak responses can be obscured by heterogeneous tissue, skin response, and background clutter. Orbital angular momentum (OAM) excitation provides structured azimuthal phase diversity, although its ability to supply recoverable lesion information beyond conventional sampling and general phase-coded excitation remains uncertain. This study presents a controlled matched-budget numerical benchmark of low-order near-field OAM excitation. A circular 16-element aperture operating from 3–8 GHz was evaluated using 160 three-dimensional heterogeneous breast-mimetic phantoms. Five prescribed azimuthal states were compared with full non-OAM, matched limited non-OAM, and matched phase-coded non-OAM acquisition. Performance was assessed using localisation error, peak-clutter ratio, target–background separability, target-hit rate, spatially constrained detection, individual-order and matched-state-set analyses, and within-method robustness to bandwidth reduction, excitation impurity, and aperture misalignment. The commanded states were correctly constructed at the discrete aperture, although this did not constitute physical-field or experimental validation. Under ideal matched subtraction, OAM increased peak-clutter ratio but did not improve spatially valid lesion recovery relative to matched limited non-OAM acquisition. No individual non-zero order or tested state configuration produced a consistent recovery advantage. Under practical background suppression, reference mismatch dominated all acquisition families. Received scattered-mode behaviour was order- and impurity-dependent but did not establish modal redistribution as a unique causal mechanism. The framework separates OAM-specific azimuthal effects from general phase coding, information budget, and coherent-control burden, and provides a numerical basis for screening acquisition concepts for subsequent full-wave modelling and physical-phantom verification.
