Scholarlink

<p>This study addresses the pressing need for precise and non-invasive skin cancer diagnostics by developing an optimized hybrid graphene-copper terahertz architecture. Despite recent advances in THz biosensing, reliably detecting and differentiating between early melanoma stages remains challenging due to the subtle dielectric contrasts of skin tissue biomarkers. To overcome this limitation, we propose a tunable graphene-assisted resonator, whose electromagnetic response is dynamically modulated via the chemical potential of graphene. The sensor shows pronounced reconfigurability, with the resonance frequency shifting from 2.55&#xa0;THz at <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\mu _c = 0~\text {eV}\)</EquationSource> </InlineEquation> to 2.96&#xa0;THz at <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\mu _c = 0.5~\text {eV}\)</EquationSource> </InlineEquation>, confirming the effectiveness of Fermi level tuning. To enhance prediction accuracy and extract hidden correlations across physical parameters, a K-Nearest Neighbors (KNN) regression model was employed. Initial sensitivity values for key melanoma biomarkers were found to be 54.5, 71.1, 75.9, and 80&#xa0;GHz/RIU. Following KNN-based optimization, these increased significantly to 87.20, 137.6, 149.3, and 187.4&#xa0;GHz/RIU, respectively, with a high prediction accuracy of <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(R^2= 0.984\)</EquationSource> </InlineEquation>. These findings underscore the decisive role of graphene’s electronic tunability in enhancing biosensor performance and demonstrate the efficacy of machine learning in guiding design optimization. A label-free and space efficient configuration is achieved through the optimized design, enabling refractive-index based terahertz sensing and offering a computationally assisted solution for early-stage melanoma detection with high diagnostic precision.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Graphene–Copper Hybrid Terahertz Biosensor with KNN Sensitivity Optimization for Early Skin Cancer (Melanoma) Detection

  • 期刊论文

  • 发表日期：

    2026年1月17日

查看全文
  • Hamza Ben Krid,
  • Hamza Wertani,
  • Aymen Hlali,
  • Hassen Zairi

摘要

This study addresses the pressing need for precise and non-invasive skin cancer diagnostics by developing an optimized hybrid graphene-copper terahertz architecture. Despite recent advances in THz biosensing, reliably detecting and differentiating between early melanoma stages remains challenging due to the subtle dielectric contrasts of skin tissue biomarkers. To overcome this limitation, we propose a tunable graphene-assisted resonator, whose electromagnetic response is dynamically modulated via the chemical potential of graphene. The sensor shows pronounced reconfigurability, with the resonance frequency shifting from 2.55 THz at \(\mu _c = 0~\text {eV}\) to 2.96 THz at \(\mu _c = 0.5~\text {eV}\) , confirming the effectiveness of Fermi level tuning. To enhance prediction accuracy and extract hidden correlations across physical parameters, a K-Nearest Neighbors (KNN) regression model was employed. Initial sensitivity values for key melanoma biomarkers were found to be 54.5, 71.1, 75.9, and 80 GHz/RIU. Following KNN-based optimization, these increased significantly to 87.20, 137.6, 149.3, and 187.4 GHz/RIU, respectively, with a high prediction accuracy of \(R^2= 0.984\) . These findings underscore the decisive role of graphene’s electronic tunability in enhancing biosensor performance and demonstrate the efficacy of machine learning in guiding design optimization. A label-free and space efficient configuration is achieved through the optimized design, enabling refractive-index based terahertz sensing and offering a computationally assisted solution for early-stage melanoma detection with high diagnostic precision.