多谱段短波红外小鼠静脉荧光观测成像研究

短波红外（short-wave infrared，SWIR）一般指900～1 700 nm的光波段，是肉眼不可见的光波段，这种波段目前主流的探测器以InGaAs为主，主要用于军事、生物以及材料光谱分析等领域。短波红外荧光成像以其对生物组织光学损伤小、成像深度大、成像信噪比高、空间和时间成像分辨率高等特点，使得基于InGaAs探测器的生物光学成像成为生物组织观测领域的研究焦点。生物光学窗口的多窗口宽谱段的荧光光谱特性，使得可以对生物组织采集多谱段的光谱图像，以此来观察生物组织的在不同光谱照明下的结构特性。针对生物光学窗口的光谱特性，设计了一种基于InGaAs探测器的多谱段的小鼠静脉成像系统，可以无接触采集小鼠静脉红外光谱图像，对测小鼠的静脉红外光谱。设计的基于InGaAs探测器的短波红外探测器，可以实现最高5 000 ms的积分时间；积分时间的延长，显著地提升了静脉成像的信噪比，同时其光谱响应特性很好的覆盖了第二生物光学窗口以及第三生物光学窗口。针对光学显微特性的成像特点以及静脉组织在图像中的特征表达，设计了一种新型的单光谱多焦距融合算法，可以很好的实现静脉图像的红外光谱观测。提出了一种基于多尺度梯度域引导滤波（gradient domain guidedfilter，GDGF）多焦距融合算法，来补偿显微特性的成像缺陷。通过多尺度梯度域引导滤波算法，实现对显微对焦区域的提取，进而实现对融合决策函数的计算，最后再通过梯度域引导滤波将得到的融合决策函数精细化处理最终得到我们的融合算法的最终决策融合函数。实验表明，设计的InGaAs短波红外探测器很好的满足了对小鼠静脉荧光光谱成像的需求，分别实现了针对小鼠静脉的1 100，1 250和1 350 nm等多个波段的光谱成像，以及在同一激光照明下实现多个焦距下的光谱成像。同时设计的融合算法很好的提取了小鼠静脉图像的对焦区域，在将多焦距图像融合的同时又减少了噪声的引入，实现了高质量全局静脉成像。

Research on Shortwave Infrared Multispectral Fluorescence Imaging of Mouse Vein

Short-wave infrared (referred to as SWIR) generally refers to the 900~1 700 nm light band, which is invisible to the naked eye. This band’s mainstream detectors are InGaAs, which are mainly used for military, biological, biological and material spectral analysis. In the field of biological tissue observation, short-wave infrared fluorescence imaging is characterized by small optical damage to biological tissues, large imaging depth, high imaging signal-to-noise ratio, and high spatial and temporal imaging resolution, making bio-optical imaging based on InGaAs detectors biological Organize research focus in the field of observation. The bio-optical window’s multi-window and wide-spectrum fluorescence spectrum characteristics allow us to collect multi-spectrum spectral images of biological tissues to observe the structural characteristics of biological tissues under different spectral illuminations, which further facilitates scientific knowledge research. In this paper, a multi-spectral imaging system of mouse vein based on InGaAs detector was designed for the spectral characteristics of the bio-optical window, which can collect the vein images of mice without contact and help observe the infrared spectrum of mouse veins. The system based on the InGaAs detector we designed can achieve an integration time of up to 5 000 ms. By extending the integration time, the signal-to-noise ratio of vein imaging is significantly improved, and the detector spectral response characteristics cover the second bio-optical window and a third bio-optical window. From the imaging characteristics of optical microscopy and the characteristic expression of vein tissue in the image, a new single-spectrum multi-focal fusion algorithm is designed to which can well realize the infrared spectrum observation of vein images. This paper proposes a novel multi-focus fusion algorithm based on a multi-scale gradient domain guided filter (GDGF) to compensate for the imaging defects of microscopic characteristics. The multi-scale gradient domain guided filter algorithm extracts the focus pixel region, and then the fusion decision function is calculated. Finally, the fusion decision function is definedby the gradient domain guided filter algorithm, and finally, the final decision fusion function of our fusion algorithm is obtained. Experiments show that the short-wave infrared InGaAs detector designed by us well meets the requirements of fluorescence imaging of mouse veins and achieves spectral imaging of multiple bands including 1 100, 1 250 and 1 350 nm for mouse veins, as well as spectral imaging in multi-focus with the same laser illumination. Meanwhile, the fusion algorithm we designed can well extract the focusing area of the mouse vein image, which can fuse the multi-focus image and reduce the introduction of noise at the same time, thus achieving high-quality global vein imaging.