基于太赫兹波谱成像的脑缺血无标记实时检测技术的研究

The fast label-free detection of the extent and degree of cerebral ischemia has been the difficulties and hotspots for precise and accurate neurosurgery. Currently available positioning qualitative methods for cerebral ischemia have the disadvantages of inaccurate, low sensitivity, and implementation complexity with unclear biophysics mechanism. Therefore, the development of a fast label-free technique for precise detection is imperative. Terahertz wave has the unique physical characteristics of fingerprint spectrum, safety and water sensitivity for biological macromolecules, cells and tissues, which make it be a precise label-free analysis technique. In this project, we propose to investigate the terahertz spectral difference characteristics of cerebral ischemic tissues, cells and molecules under different degrees. The mechanism of the label-free detection of cerebral ischemia by terahertz wave is studied. Then, the key techniques for breaking the low sensitivity and slow imaging speed can be solved using terahertz real-time imaging based on optical frequency up-conversion in wide range. By exploring the multi-spectral real-time terahertz imaging based on optical frequency up-conversion, the selection principle and method for optimal THz spectrum and their combination are established. The spectral identification and image recognition can be combined together, which will be further improved using the image fusion and pattern recognition algorithms. Thus, the extent and degree of cerebral ischemia can be detected quickly and precisely. Our proposal has the advantages of good image quality, high-speed, and high sensitivity. It will provide a new label-free imaging technique for the detection of cerebral ischemia with terahertz spectral imaging in neurosurgery.

脑组织缺血程度和边界的无标记精确快速识别是精准神经外科手术中的难点与热点。目前用于脑缺血组织定位定性的手段存在灵敏度与精准度低、实施复杂、生物物理机制不明确等问题，因此，神经外科领域亟需一种新型的脑缺血无标记精确快速检测技术。THz波对生物大分子、细胞和组织具有指纹性、安全性及水敏感性等独特的性质，具备良好的无标记检测优势。本项目拟通过研究脑缺血组织及其细胞、分子的THz波谱差异，明确THz波脑缺血无标记识别的机理；突破宽带范围内THz频率上转换实时成像探测关键技术，解决THz成像探测灵敏度低、速度慢的难题；采用THz多光谱频率上转换实时成像方法，建立最佳光谱或光谱组合选择原则和方法，将波谱识别和图像识别有机地融合起来，并结合图像融合和图谱识别算法，实现脑组织缺血程度和边界的精确快速识别。该方法具有成像质量好、速度快、灵敏度高的优点，为神经外科脑缺血无标记检测提供新型的THz波谱成像技术。

缺血性脑损伤是威胁全球人类健康的主要疾病之一，发病率和致残致死率逐年升高。目前用于缺血脑组织早期检测的方法均存在术中无法精准原位成像、耗时较长、费用昂贵等缺点。因此，亟需一种新型的脑缺血无标记精确快速识别的检测技术。本项目针对神经外科临床医学中对脑缺血程度和边界精确快速识别的需求，研究了不同缺血程度的新鲜脑缺血组织、石蜡包埋脑缺血组织、脑脊液和血清的太赫兹波谱特性，结合机器学习分类算法，实现了缺血程度的快速识别，其分类识别准确率达到了89.3％；同时，对神经细胞的太赫兹波谱特性进行了检测研究，从细胞水平阐明组织物理特性改变与太赫兹波谱的内在联系，明确了太赫兹波脑缺血无标记识别的机理；研究了不同程度冲击性脑损伤大鼠的脑脊液和血清的太赫兹波谱特性，结合机器学习算法，实现了脑损伤程度的分类识别，其识别准确率达到了96.3%。对纳秒激光泵浦的太赫兹波参量上转换探测过程进行了研究，研究了泵浦能量和晶体有效增益长度对太赫兹波探测灵敏度的影响，在铌酸锂晶体和KTP晶体中均实现了高灵敏度的太赫兹波参量上转换探测，其中，在铌酸锂晶体中实现了0.9-1.83THz频率范围的太赫兹波频率上转换，最小可探测太赫兹波能量达到了9pJ；基于KTP晶体实现了4.45THz处的太赫兹波频率上转换成像。基于太赫兹成像技术对击打性脑创伤大鼠脑组织进行多深度切片成像，采用三维重构技术，实现了正常鼠脑和创伤鼠脑的三维重构。研究了太赫兹波衰减全反射式成像技术，通过优化棱镜设计、太赫兹波偏振态等参数，建立了高灵敏、快速的太赫兹波成像检测系统，并对不同缺血程度脑组织进行成像检测，结合综合算法，对脑缺血区域进行提取，实现脑组织缺血程度与边界的精确快速识别。