面向视觉脑功能探测的高灵敏度与定量化近红外光谱成像关键方法

Insight into the brain functions is important for understanding of the brain, which will ultimately lead to the utilization, exploration and imitation of the human brain. As one of the main regions for visual information processing, the occipital area in the hindbrain gains intensive attention from the researchers in field of brain sciences. Thus far, near-infrared spectroscopic (NIRS) imaging has been developed as an advanced means for dissecting the brain functions, and offers a powerful compensation for the establised modalities. However, the currently available NIRS methodoloqies have the limitations not only in detection of the occipital region due to insuffciency of both the sensitivity and dynamic-range but also in the spatial resolution and quantitative accuracy.. The proposal herein focuses on the scientific keypoints existing in the NIRS development for brain function imaging, especially with probing the occipital area. Firstly, the fundamental elements for enhancing the NIRS proformances are addressed, e.g., the reasonably simplified layered structure for the occipital brain, the accurate photon-migration modeling and the in vivo determination of the structured optical properties, etc. Secondly, a novel lock-in photon counting mode is developed that is capable of effectively improving the detection sensitivity and measuement dynamic range while accomplishing the required temporal resolution, basd on which a prototype NIRS imaging steup of high sensitivity and large dynamic range is designed, together with a optimized scheme for the optode deployment. thirdly, an optical tomography guided semi three-dimensional diffuse optical tomography framework is proposed for significant improvement in the reconstruction quantitativness, spatial-resolution and robustness. Finally, a set of advanced strategies for hierarchical reduction of the physological or functional noises as well as for statistical inference of the event-invoked hemodynamic responses is devised, which is then integrated with the image reconstruction algorithms to evolve into a task-relevant NIRS analytical architecture.. This project aims at not only providing a novel and reliable methodology for brain functional imaging of the occipital area, but also substantially promoting the sensitivity, quantativeness and applicability of the brain functional NIRS methodology, as a whole.

脑功能研究对于认识脑并最终实现保护、开发和仿照脑具有重要意义，其中脑枕部作为主要视觉区，其功能探测在脑科学研究中占有重要地位。近红外光谱(NIRS)成像已经成为脑功能研究的先进手段，然而，目前的NIRS成像方法因检测灵敏度和动态范围限制而无法实现枕部的可靠测量，且存在空间分辨率和定量性低等缺陷。本项目针对发展面向脑枕部探测的NIRS成像方法所面临的关键科学问题展开深入研究：探讨面向NIRS成像性能改善的组织光学基础问题；发展锁相光子计数检测新原理，建立能有效提高枕部视觉功能检测可靠性的高灵敏、大动态NIRS成像测量方法；提出有效提高NIRS成像定量精度、空间分辨率和鲁棒性的光学拓扑导引半三维扩散光学层析方法；设计增强生理噪声抑制合理性和功能响应推理客观性的先进解决方案。本项目不仅为视觉功能研究提供先进的成像方法体系，也将整体促进脑功能NIRS成像方法在灵敏性、定量性和普适性上的实质进步。

近红外光谱(NIRS)成像已经成为脑功能研究的先进手段，然而，目前的NIRS成像方法因检测灵敏度和动态范围限制而无法实现枕部的可靠测量、特别缺乏用于实现高密度(HD)扩散光学层析成像(DOT)所需交叠覆盖测量，而存在空间分辨率和定量性差等缺陷。.本项目针对发展面向脑枕部探测的NIRS成像方法所面临的关键科学问题展开深入研究：探讨了面向NIRS成像性能改善的组织光学基础问题，五层平板模型和人脑MRI图谱分别发展了基于扩散方程有限元解的正向建模方法；通过发展基于方波调制策略的锁相光子计数检测新原理，建立了能有效提高枕部视觉功能检测可靠性的高灵敏、大动态NIRS成像并行测量方法和三波长(780nm、808nm、830nm)、240通道(20源×12探)脑功能NIRS成像系统；提出了一系列有效提高NIRS成像定量精度、空间分辨率和鲁棒性的NIRS-DOT先进图像重建方法，包括光学拓扑导引半三维DOT方法、稀疏正则化的脑功能HD-DOT方法、基于卡尔曼估计的脑功能HD-DOT直接重建法和面向HRF和兴奋度联合估计HD-DOT直接重建法等；在上述重建方法框架下，设计了增强生理噪声抑制合理性和功能响应推理客观性的先进解决方案；开展了有效的仿体和在体验证实验，完整地评估了方法和系统的性能和成像能力。.本项目以NIRS高灵敏并行测量技术和DOT脑成像图像重建方法为突破，以建立多波长、高密度NIRS成像系统为目标、以缜密的仿体和在体实验验证为依据，有效建立了一个理论严谨、实用高效脑功能NIRS-DOT成像体系，不仅为视觉功能研究提供先进的成像方法体系，也极大地整体促进了我国脑功能NIRS成像方法研究与仪器研发水平在提高灵敏性、定量性和普适性等重要方面进一步取得实质进步。.我们按原定计划开展了研究，并取得了预期的研究成果，完成了预期的考核指标。在第二部分将详细展示本项目的主要研究成果和进展。