为实现多肿瘤标志物联合检测的高内涵筛选，本项目提出一种太赫兹非扫描式蛋白芯片快速检测方法，将全内反射结构耦合金属纳米光栅两种宽频近场增强机制相叠加，利用砷化镓肖特基结太赫兹调制器的实时空间调制能力，实现：1）宽频非共振高倍增电场增强模式分析，建立入射场与狭缝边缘电荷激发场的波矢关系，基于完美导体近似及微观电子理论，研究倏逝波被金属狭缝分割的电场分布及近场增强效果；2）基于蛋白组学方法的多肿瘤标志物芯片设计，以电控太赫兹调制器阵列实现单探测器非扫描式大样本数据的平行采集，并以相位修正吸收截面系数及分子偶极矩的介电参数表征，解析痕量抗原超灵敏检测机制；3）利用机器学习方法建立对目标物浓度的精准预测模型，以高通量数据统计确定阳性临界值策略，结合特征提取算法解决抗原-抗体交叉反应的假阳性干扰。该项目是新型太赫兹近场探测技术的应用方法学研究，有望为基于肿瘤标志物筛选的癌症早期诊断提供有益思路和指导。

To realize high content screening (HCS) of free-label protein microarray in THz frequency, this project proposes a rapid terahertz non-scanning detection method for multi-tumor marker protein chip. Two regimes of broadband near-field enhancement mechanisms with total internal reflection and nano-metal gratings are superimposed. Particularly, research efforts are first put to analyze the mechanism for broadband non-resonant field enhancement of THz field. Based on the microscopic Drude-Lorentz model taking into account retardation processes in the metal film, the numerical relationship of incident electric field and electric field of the charged slit edges are obtained, the effects of the field distribution and enhancement is discussed by the separated evanescent wave field. Then, according to the proteomics research, the multi-tumor marker protein chip is prepared. Terahertz spatial light modulator is used to detect large samples array rapidly by one THz detector. A multilayer geometry to characterize thin film samples in reflection terahertz time domain spectroscopy with the higher sensitivity is proposed. With superb spectral resolution and robust prediction results, we expect to demonstrate the ultrasensitive detection mechanisms of antibody-antigen with trace signal. Finally, a machine learning method namely support vector regression is applied to the quantitative characterization of antigen in microfluidic devices on protein chip. The positive threshold strategy is determined by high-throughput statistics. The maximum information coefficient method is used to solve the practical problem of antigen-antibody cross-reaction, and protein chip parallel analysis and visual data processing are realized. This project, based on the notable terahertz near-field detection technology will potentially provide useful ideas and guidance for the early diagnosis of cancer based on tumor marker screening.