A high-sensitivity terahertz SPR biosensor with machine learning optimization for colorectal cancer detection

This study presents a terahertz-based surface plasmon resonance (SPR) sensor developed for colorectal cancer detection. The device employs a distinctive multi-resonator design that integrates gold, silver, and graphene. Structurally, the sensor comprises an elliptical ring resonator coated with silver, surrounded by a gold-coated circular ring on a silicon dioxide substrate, while a graphene layer is incorporated to enhance sensing performance. Performance analysis was conducted using COMSOL Multiphysics simulations under varying conditions, including graphene chemical potential, incident angles, and resonator dimensions. The proposed sensor demonstrated a maximum sensitivity of 1100 GHz/RIU across a refractive index range of 1.329–1.348 RIU, achieving an optimal figure of merit of 17.460 RIU⁻¹ at 0.719 THz. Additionally, a Random Forest Regression model was used to optimize sensor parameters, achieving up to 100 % accuracy in predicting sensor responses. The device also demonstrated potential as a 2-bit binary encoder, highlighting its versatility for both biosensing and data encoding applications.