Terahertz metamaterial-based perfect absorber biosensor with AI integration and multiband double-negative response for early non-melanoma skin cancer detection

Abstract

This paper presents a multiband terahertz metamaterial-based biosensor designed as a perfect absorber for the early detection of non-melanoma skin cancer. The proposed structure integrates a meticulously engineered multilayer architecture that achieves simultaneous negative permittivity, permeability, and refractive index (double-negative response) within the 0–5 THz range. Unlike conventional terahertz sensors that rely on one or two resonances, our design produces approximately twenty high-Q absorption peaks, with eleven exceeding 95%, four exceeding 97%, and two surpassing 99% absorption. These dense resonances enhance field localization and increase sensitivity to dielectric variations in biological tissue. Numerical simulation demonstrates a sensitivity of 629.95 THz/RIU and a figure of merit of 15,179.59 RIU−1, significantly outperforming existing metamaterial biosensors. To further improve diagnostic capability, a broadband spectral-analysis framework is incorporated to analyze the full terahertz spectral response using Euclidean distance, mean squared error, and correlation metrics. The Spectral classification framework reliably distinguishes between healthy and cancerous tissue profiles, enabling an automated and robust detection. The biosensor's performance is further validated by incorporating it into a microwave imaging system, which provides spatial confirmation of cancerous tissue. These results establish the proposed device as a compact, high-resolution, and non-invasive platform for the early diagnosis of non-melanoma skin cancer.