High-sensitivity terahertz metasurface biosensor for multi-cancer detection: a machine learningenhanced approach using graphene–MXene– silver–copper hybrid architecture

Abstract

Early cancer detection requires highly sensitive diagnostic tools beyond the capabilities of conventional imaging and biopsy methods. We present a terahertz (THz) metasurface biosensor that integrates a copper-coated H-shaped resonator with three silver rectangular resonators enclosed within an MXene circular ring. The design incorporates complex electromagnetic interactions, nonlocal effects, and coupled-mode modelling to optimise performance. The biosensor achieves a sensitivity of 1000 GHz/RIU, a quality factor of 3.6–3.747, and a figure of merit up to 13.333 RIU⁻¹. It maintains stable absorption (52.789–53.804%) across 0.27–0.281 THz, with a linear resonance–refractive-index response (R² = 0.95276). Machine-learning optimisation of graphene chemical potential further enhances predictive accuracy (R² = 0.93). By enabling simultaneous detection of multiple cancer biomarkers through frequency-shift analysis, this noninvasive platform offers strong potential for real-time, early-stage cancer screening.