皮肤组织太赫兹波段的介电模型理论及皮肤癌识别参数的研究

With the international research breakthrough and advancement of the terahertz technology for the early diagnosis of skin cancers, however, the diagnostic technology is still far from the medical practice in hospitals due to lack of reliability and the low accuracy of distinguishing the healthy tissues from the cancerous ones. Many studies have found that the tumor and healthy skin tissue have a prominent difference in the terahertz range. However, there is still lack of deep understanding of the contrast mechanism, not to mention the investigation on identification parameters. Current research is mostly focused on the dielectric models based on the water content inside the skin tissue. Such models failed to explain the more profound mechanism and the interactions of other important contents of the skin tissues. Our proposed research is to investigate the dielectric model based on the effective medium theory of multi-phases of realistic human skin tissues. The human skin tissue is classified as water, collagen, lipids and other effective components. The dielectric properties of each component in the terahertz range will be characterized individually. The cross-linking between collagen molecules, the hydration shell around the proteins will be further considered. The effective medium models will then be further developed for the multi-phases of the skin tissue based on the above mentioned factors. Moreover, phantoms will be fabricated for testing and verifying the effective medium models. By combining the experimental data measured from the skin tumors and healthy tissues with the verified theoretical models, the proposed research is to investigate the fundamental mechanism in terahertz cancer detection with the sufficient knowledge. The identification parameters of skin cancers from the healthy ones will be studied for the purpose of early diagnosis of skin cancers with the high accuracy and reliability by using terahertz technology.

国际上太赫兹技术在皮肤癌早期诊断的研究已经取得了重要突破和一定进展，但是距临床应用还相当远，主要困难在于改进皮肤癌诊断和识别的准确性和可靠性方面。大量研究发现健康皮肤组织与皮肤癌组织在太赫兹波段有显著不同的特性，但其对比机制及识别参数仍缺乏深入的研究。目前研究是对皮肤组织中水分与太赫兹波相互作用进行建模，来模拟皮肤组织在太赫兹波段的介电特性。然而，仅考虑皮肤中的水分而忽略其他成分对太赫兹信号的影响不足以深入了解肿瘤与健康组织的太赫兹成像对比机制。本项目研究基于皮肤的有效介质理论建模，将皮肤组织拆分成水分、胶原蛋白、油脂等有效成分组成的多相有效介质；将胶原蛋白的交联、水合作用进一步引入模型参数。通过制作皮肤组织仿体验证理论模型，提取有效参数；结合皮肤癌与健康皮肤组织的实验数据，深入研究皮肤癌与健康皮肤组织的对比度和识别参数。以期达到用太赫兹技术准确可靠的皮肤癌早期诊断的目的。

生物医学，尤其是关于肿瘤的早期诊断方面的研究是太赫兹技术非常重要的应用领域之一，这与太赫兹波的几大重要特性有关。首先，氢键的振动能级处于太赫兹范围，而液相水中含有大量的氢键，因此，太赫兹波所处的频段对水分的变化异常敏感。而肿瘤组织往往表面血管分布更加密集，从而导致肿瘤组织的水分含量高于健康组织，这为早期太赫兹癌症诊断技术提供了依据。其次，太赫兹波的光子能量约为4.13meV，因此与传统的医学成像手段（如X光）相比，太赫兹不会对人体造成电离伤害，是一种安全的弱电磁波。研究表明，其他波段的光学成像方法在肿瘤组织与健康组织的对比度上相对弱于太赫兹成像，然而产生这种对比度的机制尚不明确。本项目通过对生物组织中各主要成分的太赫兹谱进行测量、分析和仿真，明确了脂肪酸等生物分子与太赫兹波相互作用的机制；通过图像对准、融合以及机器学习的方法，对健康组织与癌症组织的太赫兹图像进行自动分类，初步结果显示，对于病理样本的识别准确率达到80%以上。